 Yeah, thanks. Thanks a lot. And then we move quickly to the last paper and I would in mind present a discussant to be also very sharp in time because we are a little bit behind. This paper will be presented by Anastautowicz from the ECB. Thanks to the organizer for reminding us here to the conference and it's a pleasure to be here and thanks also to the scientific committee for selecting our paper. I will be presenting a joint work with Leonardo Gamba Corta from the BIS and Alessio Reggietta from the ECB. The paper is entitled the supervisory policy stimulus evidence from the Euraira and recommendation. And yes, still disclaimer for me. The views expressed here in this paper are of course of the authors and do not necessarily reflect the BIS or the ECB views. It's not full screen. I don't know if it has to be full screen, but anyway. The COVID-19 pandemic has prompted governments and central banks to implement innovative policy solutions. One of the main innovation was the introduction of the policy to restrict dividend distributions. In Europe on 27th March 2020, the banking supervision arm of the ECB adopted the recommendation asking but not obliging banks to refrain from dividend distributions. The objective of the recommendation was to ensure that credit institutions will continue to fund the small and medium enterprises and strengthen their capacity to absorb losses. To our knowledge, this is an unprecedented policy. It is the first type of supervisory policy that restricts dividends that at least recommends banks to not pay out dividends. A bit of the literature background, in restricting dividend payments, there are of course caveats to be kept in mind. Public policy must be crucial not to create unintended additional distortions that exceed the benefits of restrictions. It is worth pointing out that dividend restrictions can have also negative effects on banks. The first one that comes to mind, of course, is the stock price effect. And there is evidence on that. Those negative effects are generally, however short lived. The stock price is still maintained. A mean reverting behavior. And this can suggest that initial market reaction can be excessive. At the same time, the idea of restricting dividend payments to support lending is not new. Literature has already argued in that front and argued for banking sector-wide dividend restrictions in crisis. While dividends represent the legitimate distribution of profits to shareholders and normally play an important role of allocation of capital between savings and investments, empirical evidence suggests that in times of crisis banks tend to expand dividend distributions instead of reducing them. This is this as a signaling device, but this also implies that they act prosyclically. Finding a balance is not easy, of course. And in crisis, however, supervised intervention can be justified on those fronts. The next slide shows you the dividend capital allocation of the banks, managers, and some insights on what we see in the data. Bank managers face a choice of capital allocation when deciding to follow the ECB recommendation. On the one hand, they may talk to use surplus capital to increase lending supply, thus supporting the economy and acting consyclically. This seems to be confirmed in our data. The chart shows you the spike of not distributed dividends following the ECB recommendation and the spike in lending in March after March 2020. On the one hand, bank managers, however, can still decide to signal and increase their resilience to future shocks by saving capital or strengthening their loss of social capacity, setting aside low-loss provisions. In short, the impact of the dividend recommendation on lending is not given and different things have to be considered. What we do in this study, we try to estimate the impact on lending growth to non-financial corporates, first and foremost, and then we try to investigate also how these credit allocation across firms have changed. We look at vulnerable and invulnerable sectors in the economy and those more affected or less affected by COVID. Finally, we look at the riskier behavior by banks and if there is any. Our data, the dividend data that we have are coming from the SSM or ECB supervisory banking supervision survey that was held, that was conducted in March 2020. The chart shows the aggregate figures from dividend distribution plans and intuitively shows that the banks payout ratio out of fiscal year 2019 earnings is 45%. However, if the condition on banks that have positive dividend plans, this ratio goes to 57%. What can be seen is that the middle yellow bar is the 27.6 billion of 2019 dividend retentions. These dividends are the ones that are the focus of our paper and that were a bit held following the ECB recommendation. In terms of regulatory capital, this amount could be read as an additional release of 33 basis points of CAT1 capital that banks can use to provide credit supply, capital increase, or loan loss provisions. Identification. So this slide briefly guides you through the identification that we exploit in this paper. Our main variable of interest is the non-dividend, not distributed dividends planned, but not distributed dividends over risk-related assets. This is basically the measure of the intensity of treatment. The chart shows that the distribution of this variable in the baseline sample and there is a spike at zero. It's around 40% of observations which refer to our control group. This is formed by two types of banks, banks that did not have any plan to distribute dividends and banks that distributed all that they planned to distribute before the March 2020 recommendation. Then we have the remaining 60% of observations that are distributed across this chart and they refer to our treatment group. Those are the banks that follow the ECB and did not distribute their dividends. For identification, we then rely on different features, but the most important ones is that we take the survey prior to the ECB recommendation that took place, which means that self-selection issues and the genetic issues that may follow after the period are kept in check or at least the biases in our estimates can be kept in check. And of course, the COVID pandemic is widely recognized as exogenous shock to the economy and therefore also the dividend recommendation by the ECB could not have been anticipated by the banks. Then secondly, the decision on dividends is predetermined to the COVID shock and to the dividend recommendation and the dividend recommendation will not affect those. And finally, the econometric framework, we rely on credit registry data to isolate firm demand effects. And to this end, we use the unaccredited credit registry data of the Euro area. And this allows to control for an unobserved heterogeneity credit demand at firm level. And in extreme synthesis, we do what the rest of the literature does here is comparing two banks lending to the same firm with banks being differently affected by the dividend recommendation, some of them being more and some of them less. To do a correct estimate, or at least to try to do a correct estimate, we need to focus also on the confounding effects that are happening at the same time, different type of policies, fiscal monetary potential policies have been implemented at the same time of the dividend recommendation. So keeping in check those is not easy. We try to do our best by collecting the data that we have here at disposal at ECB and we focus in particular on fiscal policy measures, monetary policy measures, and potential policies. On the left hand side chart, with respect to fiscal policies, we control for government guarantees, the blue line, moratorium debt repayments, the red line, which both aimed at support credit during the COVID crisis. As you can see, all of those variables spike at the same time. We then include in the model measures of unconventional monetary policies, the asset purchase programs and the TLTROs, which at the same time as well were enacted by the ECB. On the right hand side chart instead shows the fall in off-balance sheets, off-balance sheet exposures, which are mostly credit lines in blue. As you might know, credit line drawdowns were one of the main sources of credit for NFCs during the early phase of the pandemic, and when they are drawn, they are automatically moved to balance sheet, increasing lending mechanically. We add in the second control set order to measures of potential policies, namely the combined buffer requirement and the P2G releases, which have been touched upon earlier. This chart shows the first results of our paper. In columns one and two, we have the baseline, where we see that the positive relationship between non-distributed dividends and the credit growth. More specifically, one percentage point increase of non-distributed dividends over risk-weighted assets is associated with additional lending growth of 4.3 percentage points. To understand the economic significance of these effects, we note that conditional on treated banks, the average ratio of non-distributed dividends over risk-weighted assets is half percentage points. So we can, as a rule of thumb, halve those estimates to see how much more lending was provided during this period by those banks. And we come up with the 2.1, 2.2 percentage points of higher lending by treated banks with respect to the controls. In column three and four, we assess whether the supervisory dividend stimulus is directed to buy micro, small, and medium enterprises. From a policymaker's perspective, it is important to ensure that those firms are provided funding in a crisis, in particular, because those are the ones that have higher difficulties accessing credit and they cannot rely on financial markets to fund themselves, or usually do not rely on financial markets to fund themselves. We find that the impacts are strong for those small and medium enterprises, however, lending growth to micro enterprises is lower, and significantly so. This suggests that micro enterprises can be perceived as riskier or more vulnerable during periods of systemic shocks, in line with theories of packing order that have been recently published in the literature. In column five to six, we focus instead on vulnerable sectors, and we look at the firms and sectors that were most affected in terms of revenue by the COVID-19 lockdowns, and we see that the impacts are higher for actually for those firms, and significantly so with respect to the firms that were not. This suggests that dividend restrictions may have helped also some firms to survive the COVID shock, which will not arise. We don't have evidence on that, but these findings point to that direction. The next slide, I would like to show you some interactions with government guarantees. The results reported in column one and two are important for two reasons. First, in the first row, the coefficient of non-government guaranteed loans is positive, sizable and statistically significant, even though at only 10 percent level, indicating that credit supply grew independently of the government support. Second, we find strong complementarity with this fiscal policy measure as the interaction term flows, and most of our effects are driven by those guaranteed loans. Column three and four instead focus a bit on the distance to MDA, which was also explained earlier in the previous presentation, and it's the main measure of capital space the bank has a disposal to distribute more loans. Point estimates indicate that most of our results are explained by banks with higher capital space. We think that these finding points to the possibility that banks close to the MDA trigger are likely to have used the non-distributed dividends to build up much needed capital instead of lending. The last results that I would like to show looks at the risk-taking by banks. We define zombie firms as those firms above the 95th percentile of accumulated impairments within a bank-firm relationship, and as of 2019 Q4 to avoid endogeneity effects. Lending to such zombie firms will call for a type of gambling for resurrection. Excessive risk-taking by banks of this type could result in additional not needed losses. The results, however, indicate that the increase in lending supply is not driven by those firms with substantial impairments. If anything, lending to those firms is lower as Column six shows, and in particular the marginal effect of lending to more problematic zombie borrowers is not statistically different from zero as shown from columns three to four to six. In the last two columns, we turn to MPL ratios and banks with structurally high MPLs. We see that those banks do not take on additional risks. Actually, these banks take on less risks than most of the loans are provided by banks with low MPL ratios. Finally, let me turn to some conclusions and policy implications. Given the restrictions, we hope, after showing this evidence in the paper, can be an effective policy to sustain credit in a crisis in the downturn. A proven counter-cyclical supervisory policy that was the first time that was enacted. But possibly, this can be used also in financial crises, too. Of course, it depends on the type of crisis. Given the restrictions can enforce the effectiveness of counter-cyclical policy, and they can work together with other type of policies, fiscal policies, monetary policies, or other type of financial policies. The second point we'd like to make is that given restrictions can effectively more financial resources from inefficiently high shareholder consumptions to credit supply, our results show that non-distributed dividends are channeled to loans which are likely to have a higher growth multiplier than consumptions of shareholders, in particular in the downturn. And we note that at the same time, banks could benefit from this additional credit through higher net interest income. Few more aspects we think are worth highlighting. A caveat is due. The temporary nature of dividend restriction is however necessary to limit unintended policy effects, so strong components of supervisory forward guidance should accompany them. Clear communication on the duration, their rationale, and their implementation should help to limit inefficiencies owing to uncertain policy environments. Otherwise, financial stability can quickly be undermined. Additional welfare improvements resulting from dividend restrictions are worth pointing out. Dividend restrictions can increase solvency and loss of social capacity at the same time simultaneously. This is a feature that capital releases might not have. This means that in the event of a bail-in, debt holders and potentially also tax payers will bear a lower cost. Moreover, dividend restriction can address one of the problems related to the releases of macro potential or general capital requirements, which can be misused to distribute more dividends instead of to provide more loans or set aside capital. By combining the dividend recommendation or one type of some sort of dividend restrictions with releases, this unintended effect can be cancelled. We think that overall, based on our paper, we have a new kid on the block and there is a new policy tool at disposal of supervisors to be used in the future crisis. Thank you very much for your attention. Thanks Ernest and the discussion is Diana Bonfin from Banco de Portugal and Katholica Lisbon. Thank you and the ECB as well. I'm in the opposite direction of Florida and I get more affiliations and actually, yeah, the disclaimer is so long that I think it's not even here on the slide, but it for sure applies. Okay, so basically, I think what the paper tells us about is this restriction on dividends that was implemented at the beginning of COVID. So if we think about how things work in the crisis, so providers want to be sure that the banks have the ability to absorb losses, right? But if you if during the crisis, you're going to ask banks to increase capital requirements, this is going to have prosecutical effects. And so one solution that was adopted here two, three years ago was to tell banks not to distribute dividends to make sure they would be able to increase their loss absorbing capacity. So the recommendation issued at the beginning of the pandemic said that at least until October 2020, no significant institutions should pay out dividends. And so what Ernest and co-authors doing this paper is to try to understand how did this recommendation issued by the ECB affect banks' lending and risk taking. So they find that this dividend restriction was actually quite effective in avoiding a credit crunch. So what the banks did was to use these non-distributed dividends to lend more and to lend more especially to smaller and medium firms, as well as to firms in the sectors that were being more heavily affected by the pandemic. And they also cannot find any evidence of additional risk taking or zombie lending. So my comments will be mainly on three topics on bank profitability and dividend bonds, on why do banks, understanding why do banks follow the recommendation and then I will also talk a little bit about the magnitudes and finally also a little bit about the interaction with other policies. So the identification strategy here in the paper is to explore heterogeneity across banks in the area on the dividends that the banks had planned to distribute but did not do so. And this is reported as a percentage of risk weighted assets according to a survey that the ECB conducted in March 2020. And so when we think about this treatment variable, higher values of planned but not distributed dividends can mean different things. Well first they can mean compliance with the recommendation and this is the interpretation of the authors. So if I see that the bank was planning to distribute dividends but it did not well the bank was for sure compliant. But it can mean other things. It can also be related to risk preferences of the bank managers. So this is a soft recommendation so the banks could choose whether or not to comply and we cannot be totally sure that the bankers wouldn't decide to do this even if the ECB had not recommended them to do so. And actually I mean if you look at how banks were behaving at that time they also increased their impairments. So this would be consistent with this evidence of prudent behavior from the banks early in the pandemic. So the banks were claiming they wanted to be part of the solution not part of the problem this time so maybe they would have restricted the distribution of dividends anyway. And another thing that this variable can be related to is bank profitability. So the banks that had higher amounts of dividends to be distributed were for sure those that were more profitable before COVID started. So another possibility looking into these results could be that well maybe the banks that were more profitable before the pandemic were in better conditions to then lend more once the pandemic starts. So I mean here just to conclude this part I think it would be important to discuss these alternative explanations for what we see and how we look into the data. But I'll come back to this. My second set of comments is about why do banks follow the recommendation? So first I mean by when the bank managers decide to follow the recommendation there's two different things at least that they can do. So in the slides over three but here I aggregated them in two. So one thing the banks can do is well let's act counter cyclically and increase lending. And this is the goal of macro potential authorities right so so macro potential policy would want to kind of have this counter cyclical dimension and make sure that at the peak of the crisis there's not going to be a credit crunch. But banks can also decide to follow the recommendation and not distribute dividends to increase their loss of sovereign capacity and this would be more the goal of the micro potential authorities of the bank supervisors. And what we see in this paper is that the decision that the ECB adopted from a macro potential perspective actually had very positive macro potential effects. So I think this is great news for macro potential policy and I think it also delivers food for thought in terms of the interactions and potential conflicts of interest between macro and micro proof. But where I think the paper falls short a little bit is actually showing us evidence about the primary goal of the supervisors when announcing this. So there's this technical report that I participated in that where there's some descriptive evidence about what happened to banks capital and what did they do how did they adjust the balance sheets to as a reaction to this recommendation. But here in this paper the focus is really on the lending activity. So I think it would be important to complement the evidence in this paper also to this broader adjustment in balance sheets and to understand how these different dimensions interact. Still on this point I think the empirical strategy in the paper is not straightforward. So the authors start with these 110 banks and there were 35 banks that in the survey in March 2020 they said immediately they were not planning to distribute dividends. These go into the control group and then there were 75 banks that said they were planning to do so. And among these there were 53 that even though they were planning to do they didn't do it and so this is the treatment group. This is where the action is being explored. And then there are other groups that there are other banks that go into the control group. There's 11 banks that have already distributed or approved it. There's one bank that actually distributed more than planned and there's 10 of the banks that went ahead and distributed what they planned. And so in the paper there's some discussion about how to think about this control group. I think much more of it is needed because this is clearly not an easy discussion. Actually something where this is also important is to have more convincing evidence about parallel trends because these banks are not the same. As I said already these banks had different profitability levels. So banks that were planning to distribute more dividends for sure they were more profitable. And given also that compliance is a choice of the banks I think it would be really important to be clear first on the discussion on how the control group is formed and second on showing us more evidence about parallel trends. For instance, Jose Luis showed us these charts with the dynamic differences and differences that I think is a much clearer way to understand how similar these banks were before the recommendation. Also all the results are anchored on this intensity of treatment by exploring the differences in the amount of dividends that the banks plan to distribute. I think also just looking into treated versus non-treated banks which is less linked to profitability I think this could be something worth exploring. Then I have one question about the magnitudes that are reported in the paper. So the main result the number that you want to take away is that in the absence of this ECB recommendation, lending growth would have been 2.1 percentage points lower. But then a little bit later in the paper when the authors go and look into the role of government guarantees on bank loans they show that this effect in the absence of government guarantees would have been between 1.5 to 1.9 percentage points which is almost the same as very close to the entire effect. So I wonder if this meant that actually the dividend recommendation had much stronger effects on bank lending than the government guarantees. Of course it's not easy to disentangle this but I think it's worthwhile trying to address the question. Also the results are much larger for firms that borrow from several banks at the same time. So here looking into the magnitudes of the effect I find this puzzling because another conclusion in another part of the paper is that the effects are actually larger for small and medium firms which are less likely to borrow from multiple banks. So again this is something I would explore more. Then my last point is about other policies okay and of course this is a tricky point an easy point to make rather than and not so easy to address. I think the authors do a great job in trying to control for as much confounding policies that were implemented at the time. So on fiscal policy they control for the information that exists in an accredited which is government guarantees and moratorium loans but here on the right hand side this is a picture from the ESRB report on fiscal support during COVID and for instance that first set of bars is about other fiscal policy measures excluding the government guarantees and so that is also a big element of support and that cannot be captured here. Also monetary policy is captured by the uptake of TLTRO3 and the deposits at the central bank I wonder if this is enough and also we don't know enough about the buffer releases. Then just one minor thing I mean so the paper claims to be the first to this to be the first time that dividends restrictions were implemented well actually in Portugal they were implemented I don't remember if it was under government constancy or right after that sorry for that I should have but this was this was when we were trying to avoid the bailout. So before the strike it came in one of the decisions that was implemented was first let's ask the banks to stop paying their dividends and so well maybe another paper needs to be written on this to get more external validity and but so just to conclude I think that sorry just to conclude it seems that I mean the results in this paper are very clear in showing that recommending banks to restrict dividend distribution seems to be quite effective and does not lead to distortions so these are great news but it's actually kind of surprising there aren't any drawbacks so I wonder if this is the the silver bullet that now we have to solve any kind of problems okay so thank you so much. Thanks a lot Diana and now we have time for our questions first from within the room. Thank you I'm Jean-Doir Collier from HEC Paris. I just wanted to follow up on Diana's last slide actually so what seems a bit special with this with this policy is that the dividends were already planned and then you cancel them right so if you take the signalling theory of dividends seriously what this means is that you already benefited from the signal and then you don't have to distribute the money so of course this has a positive impact on lending in a sense right you get it for free even. I think that the real cost of such a policy would be that if you do it you know repeatedly then you lose the signalling function and presumably you have issues with the allocation of capital among banks and so I was wondering whether there was a possibility to look at that you know measures of how capital was allocated in the banking sector because that's where I would look for potential costs thank you. Thank you for the question any other questions from the floor so that's the case Andrea something in the chat. Let's wait a few seconds but so far everyone is satisfied. Okay then I will take another question and then if you have less questions that's also fine to bring us back to the schedule. You looked very well at the positive impact the intended impacts of the dividend recommendation and I wonder to get the full picture wouldn't you also need to look at that is the funding cost of banks also on a more persistent basis if there was also maybe a price to pay thank you. So first of all thank you Diana for very constructive comments and a lot of them actually so we will try to go through your slides and improve the paper. Regarding the first one in terms of the risk preferences of managers yes of course I mean we don't have a counterfactual on that side of course because it's going to be difficult to to see what will have happened with those dividends I mean my sense is that the banks will have distributed them anyway perhaps maybe a couple of outliers not I don't think that it could have been systemic systematic behavior by banks to not distribute dividends given the COVID crisis but still it's a valid point and maybe we have to touch upon that in the paper to make sure that this at least has been suggested by a by a discussion. Profitability we have a table in the paper that first of all we control for a bunch of bank-level variables profitability is among them some indicators of profitability are among them and then we have a table in the in the annex that looks only at banks that had positive plans to distribute dividends and not and exclude the banks that from the control group the banks that did not that are generally less profitable so they don't have anything to distribute to the shareholders and actually out the point estimates are the same weirdly enough the same in size and direction to the baseline but we can of course reinforce that argument in the paper. Interaction between micro and macro of course it's it's all capital requirements you know macro macro potential capital requirements are the same type of they are just called differently but in the end it's it's they have different reasoning of course and different rational why they are applied but it's still you know capital over its created assets so there is an interaction over there in terms of micro and macro it's it's nice it's nice that you pointed out to that and we will we will flag it as well maybe the technical report on capital provisions and at the bank level this was at the bank level and at the bank level we could of course observe you know the capital behavior how much of these funds were allocated to capital how much of two loans and how much to two provisions we done a credit we go a bit deeper to to reinforce our our identification of the effects and we cannot really look into the into one of the aspects in terms of capital there is this technical report over there we didn't we don't think about expanding into paper but we mention it in the we we indulge in self citation and we mentioned in the paper as as as evidence of some of the alternative alternative distribution mechanisms or allocation mechanisms of this of this policy then regarding your sample ish of 110 versus 99 or 100 so 110 is the is the total banks of of that we have in our sample but then in our estimations we have 99 because of missing data and other things that that affect this we take on board the the suggestion of the dynamic difference in difference we will implement that and try to see how that that what what does that tell us us and discreet discreet treatment variable as well so thank you very much for for those points we need to rephrase a couple of sentences i i i realize in the in the absence of government guarantees because there is a confusion there and what we say that we find a an impact of 2.1 percentage points if we scale back you know the the classical interpretation of increase of 100 percentage points the impact on estimates we have in our sample banks that have this ratio that we use as the main variable being around half so if we have the effects we get to 2.1 without government guarantees it's still a 1 percent to interpretation so it's not it will be helped also the other one so that's that's that's the thing and thank you for pointing out so this wall street journal article we we we missed that one and that's actually very useful and we will look if there is anything that has been written about that and actually what banks did at that time if they followed it or not so what what was the behavior um general thank you for your comment um of course here it's kind of as you say it's it's an easy choice we look at the landing and of course that you're going to find some positive effects the question is how big and you know you want to do it properly and everything you to see um in terms of costs there are costs and this goes back also to the question by by klaus there are costs that we don't address directly in this paper that are difficult to address also in terms of cost of funding because at that time in 2020 no banks basically issued any capital I mean it will have been a strange choice to issue some capital at the time of course because the stocks were plunging and you know the cost of funding of course was higher but in the medium term and actually coming back from from a recent conference in Brussels we were thinking about the medium-term cost of these policies so you can look at for example at the issuance is before covid and issuance is now in the last couple of years and how you know the pricing or the the the demand versus the offer the supply and demand effects have changed in respect to the how much banks were defund demanding in terms of capital how much was the supply there what was the different cost it's it's going to be challenged to collect the data but I think that that avenue can can can answer some of those questions in particular comparing those banks that decided to strictly follow the the ECB or comparing also actually the US with Europe the US they have also different type of recommendation but not as strict as as in Europe so it was more kind of rule-based instead of to court so maybe that could be that could be an avenue for future research of course to to to be looking at the medium-term cost of these of these policies and I agree with you it cannot be repeated it cannot be a repeated interaction game between the supervisor and the markets of the banks because of course financial stability and and other issues can be can be undermined this also goes back to to your question clouds thank you very much yeah thanks a lot so I think we have all earned now a short break so we reconvene at 1600 4 p.m. and let's give a hand to the great presentations the great presenters and also the good questions that we have as you have been asked