 So thank you very much for the invitation. Delighted to be with you, even though it's only online. Although I see at least one person in the audience, I think in three days in Zurich, if I understand this correctly. So this is a very different paper from Chef Nam, whose presentation I very much enjoyed. It's a more focused paper, more narrowed actually, because we focus here specifically on the banking sector. And specifically we focus on the effect of COVID-19 on the shock and everything that came with it on the banking sector in the US. The original title actually was, Our Banks Catching Corona. Obviously then, we've been working on this paper for almost a year now. We have changed the title, Have Banks Caught Corona. I'm not sure whether the next iteration will have yet another title. Then we just mentioned this is joint work with Jan Keil from Humboldt University in Berlin. So as I mentioned, look at the effect of the pandemic and lockdown on bank health and also bank lending. Now, if you think about, if you kind of want to frame this a bit on a theoretical level or conceptual level at least, typically what we think about what happens during a recession or a crisis is that there will be a lending retrenchment, right? For all kinds of reasons, dropping collateral values, increasing agency problems, bank losses, funding problems for banks. However, this hasn't really always happened like this, including during the global financial crisis, there was initially an increase in CNI loans on banks balance sheets, mainly due to the drawdown of credit lines and has been shown during this COVID shock. This actually has also been the case in the initial phase of the crisis. It's a bit clearer when it comes to loan conditionality. I think there's a clear indication that during a crisis, during the recession, the loan conditionality typically tightens. Now, there are a lot tons of papers already, I would argue, on the COVID crisis and the reaction or the impact on the banking sector. I mean, there has been a bit of a productivity shock for a positive productivity shock for us, especially last year for us economists. Now, you wonder what is different in this paper? And guess what we are looking at or what makes this somewhat different is that we look at how the exposure to the pandemic on the bank level and the exposure to the lockdowns has affected banks' financial health and banks' behavior. And there we look at the U.S., not just because everybody looks at the U.S., but also because there we have this variation, as I'm gonna show you here right in the next graph, we have this variation in terms of how the pandemic has spread over time across different parts of the U.S. So this starts here from the first down to the fourth quarter of 2020, all data after 2020. Plus, also there have been, of course, marked geographic differences in the use of lockdown policies, which of course have an additional economic, constituting additional economic shock for the areas concerned. And this, and I'm gonna show you in a moment that this is also, it's just not what the eyes tell you, but also in the regression, there has been also a correlation of both pandemic and lockdown independently with unemployment rates across countries, across states in the U.S. So how do we use this information in terms of the pandemic and in terms of the lockdown? So we look at, we take the branch network of banks plus the deposit distribution across banks, across branches of each bank to kind of construct a time-varying exposure measure of banks to both the COVID, measured by COVID-19 deaths per capita on the county level and to lockdown measures on the state level because there hasn't been much variation within states in terms of lockdown, at least not in most states, to kind of construct this exposure matter on the bank level in terms of how much they have been expunged to the pandemic and to the NPIs. And to give you two examples here, the red one is Citibank, which has been exposed much more to the COVID-19 crisis, given where the locations of their branches are and also their depositor base, as opposed to Xeon's bank core, which is headquarters in Utah, and has mostly branches and therefore also depositors in areas which have been less exposed to COVID-19 and the lockdown measures. Now you wonder, depositor distribution doesn't actually match with borrower distribution, it starts actually. So we do have a couple of alternative tests that we use borrower distribution according to HMDMA and according to PPP, not PPP, sorry, the Small Business Loans, and they get very similar results in terms of the geographic distribution of banks and therefore they are then also exposure to the pandemic and the lockdown measures. So, but first of all, let me take again a step back and just show you again the areas which were more affected by the pandemic over time. And also more to the lockdown measures. So these are the red ones above median of COVID deaths. Red here again in the bottom thing, the top 12 tile in terms of lockdown measures, they also experience a much higher increase in unemployment. Lockdown measure basically is this measure, I'm sorry, the name just doesn't come to my mind, Olivier, Olivier something, sorry. A very smart French economist who constructed this very interesting lockdown measure of both across countries, but in also across states within the US ranging from zero, no lockdown to six, nobody can leave the house. Right, so this is kind of more as a kind of motivation or kind of construction of the data. And so the first step what we then do with these data is we look at the effect on loan loss provision and non-performing loans regressed on these, both on the time trends, but also on the bank's exposure to pandemic and to the lockdown. The first thing you can see, this is just average data. Yes, there was an increase in both loan loss provisions and NPLs, more so in provisions than in NPLs, although also then later on it was reversed both in terms of non-performing loans, but also in terms of the loan loss provisions. Now, if you exploit the variation, the one thing you say is what we do in all of these regressions, we have again data on the bank quarter level. We have data from 2019 and 2020. Our benchmark quarter is the fourth quarter of 2020, sorry, 2019, so we have data for 2019 and 2020. Benchmark is the fourth quarter of 2019. And so if you look at these four quarterly dummies for 2020, you kind of pick up the national trend, nationwide trend in terms of loan loss provision, in terms of non-performing loans, which is the increase, more so loan loss provisions than non-performing loans. But in addition, we also find this direct link between the exposure of banks to the pandemic, again geographically over time, and to lockdown measures, both in the case of loan loss provisions, only in terms of lockdown measures in the case of non-performing loans. So in general, it's much harder to explain variation in non-performing loans across banks with exposure based both national trends and exposure to a pandemic than it is in the case of loan loss provisions. This is kind of the first result, financial health. So yes, there was a deterioration in financial health for banks more exposed geographically to the pandemic, to lockdown measures, more robust for loan loss provisions, not surprising than for non-performing loans. As we know, of course, there was a relaxation of the rules for classifying loans while banks at the same time were encouraged to actually also increase their general provisions during the crisis. So the next thing we do is we look at lending behavior. How have banks reacted to the pandemic, to the lockdown measures in terms of lending? And they are very different results, depending on whether you include the so-called PPP loans, Paycheck Protection Program, or not. These are the average numbers here. So the continuous line includes all the loans, including these government-supported, government subsidized loans to small businesses. I think it's up to 500 employees. And if they do not fire people or if they rehire people, they don't have to pay back the loans. It's partly actually a grant. It's even more contrasting for small business loans, where you have an increase in overall lending if you take into account these PPP loans, but actually without the PPP loans, lending to small businesses actually went down. You can alternatively say, banks replaced regular lending with PPP loans during the pandemic. And that also shows up when we actually look at the regression context. So first, this is overall loans and leases in the U.S. again, across banks over time. The general trends actually, it was rather negative, but more so when we exclude PPP loans. There's some variation here in overall lending, which increased for banks that were more exposed to these lockdown measures. That becomes again very striking, is when we consider the small business loans. First on the left, on the right-hand side, banks and it was a general trend to reduce regular lending and even more so for banks that were more exposed to pandemic and to lockdown measures. But at the same time, there was an increase in overall lending and this was driven primarily by banks and more exposed to the lockdown measures. Again, using these PPP loans rather than regular lending and therefore also responding to a need by enterprises that had liquidity needs, additional liquidity needs during the lockdown measures and the economic consequences. This is actually also quite a big results. So for example, if you look at the average bank, like the bank that was exposed to the average to the mean to the pandemic and to the lockdown measures, overall there was an increase of 30 percentage points in lending growth, including PPP, but it was a reduction by 10 percentage points if we exclude PPP. So of course the question is, is it really, is this demand driven, is it supply driven? Why is it that banks reduce general lending? Well, that's something we're gonna stress in the next step. Let me first show you that if you look at household loans, this did not hold, there was actually a general reduction in household lending. For CNI loans, again, a general increase in lending but not varying across banks according to the exposure, geographic exposure to the pandemic and to lockdown measures. Maybe as a final remark here, why do we find such a clear significance for geographic exposure to the pandemic and lockdown measure when it comes to small business lending but not general lending? Well, I guess this is another kind of piece of evidence that distance still matters, including during the pandemic. I mean, there's also other evidence that smaller banks and more kind of local banks were much quicker to push out the PPP loans to their clients than larger banks that might not necessarily lend to clients close by. So, how do we address the demand versus supply? So here what we do is we use data from the small business administration. So where the data have to be reported to the small business administration under the, under legal requirements. So these are data where we know how much each bank has lent to small businesses in which area, in which county, and we can compare the lending under the PPP, the paycheck protection program, with the small business lending, which in this case we averaged over 2018 and 19. Now, of course, these SBA lending and PPP lending are not completely comparable because the requirements are somewhat different, but it seems to be a very close match in terms of the target group of small business lending captured on the SBA and the PPP lending that was supported by the federal government in 2020. And what we have seen, so what we can compare here is to which extent is it local circumstances, so exposure to pandemic, exposure to lockdown measures, to which extent is it the exposure of banks on their national level to pandemic and to lockdown measures that can explain increase or decrease in small business lending under the PPP compared to pre-tendemic regular small business lending. And what we can see is that actually, it seems that the pandemic itself doesn't come in significantly. Yes, in counties which were more affected by lockdowns, there was a increase in small business lending, but also, and even if you can control for county fixed effects, the banks that were exposed more on the national level to the lockdown measures, they also lend more to small businesses in given counties. So basically banks that were more affected, again as I showed you earlier in terms of financial health, by being exposed to lockdown measures, they were more likely to replace regular small business lending with PPP lending, a kind of indirectly taking advantage of the subsidized lending to kind of support their own loan portfolio. Which maybe is an additional new insight that people haven't talked much about yet. So this is what's kind of the second part. Let me go to the third part. So I've kind of looked first at the financial health. Then I looked at lending, but of course when we look at this lending, especially small business lending, of course there's a mix of kind of the regular market-based lending or bank lending, regular bank lending, sorry, and the kind of government supported lending. Now, if you turn to a segment of the borrow population which did not directly benefit from government support, that would be the syndicated lending market. So very big loans syndicated across different banks, though of course there might also be indirect support, but it definitely not direct support as under the PPP. And so here we might be more clearly picking up effects of the crisis and both the health crisis and the economic crisis on this lending segment. So the first thing we do here is we look at the number of loans given by banks, again, a time for natural variation and then also direct exposure. Unfortunately here we have only data until under the second quarter because then the database where we got the data form doesn't have any deals can, doesn't include these data any longer. But we clearly see is that banks that were given their geographic distribution were more exposed to the pandemic that entered the lockdown measures were give fewer loans and the average loan volume also reduced. And that's in addition to the kind of national trends. When we then look at loan conditionality, so comparing looking at interest rates, looking at maturities, here is the kind of graphic illustration. Yes, banks that were more exposed geographically, in this case to the pandemic, they also increased their interest rate spread more and somewhat at least decrease also their maturity of the loans. And that's also confirmed in the regression analysis, although more so for interest spreads than for maturities. So we can clearly see here an increase in the interest spreads overall on the national trends but also for banks that were more exposed to the pandemic and to lockdown measures. Again, if you do this for maturity, so here, yeah, it's not as clear. Again, there's national trends to show the maturities, but it's, if you look at the bank specific exposure to the pandemic and to the lockdown measures, it's not as clear cut. So let me summarize. So what we find in this analysis for the US is that banks geographically more exposed to the pandemic and lockdown measures shown increase in loan provisions and non-performing loans. And by the way, in case this question might come up, we don't find similar trends in terms of deposits or liquidity or other measures. So it's really about financial health and it's also about the lending side. But we find especially an increase in lending to small businesses for banks more exposed to the pandemic and lockdown measures. So this is driven by government guaranteed loans. So government guaranteed loans basically replaced regular loans, two small businesses, four banks that were more exposed to the pandemic and lockdown measures. And similarly, we find an increase in interest spreads and decrease in loan maturity, again, which is in line with kind of standard predictions that you would have of how banks react during such a crisis or during an economic crisis when collateral values drop and the agency problems between borrow and lender increase. And let me stop here. Thank you. Many thanks, Thorsten. I believe that the index that you are referring to is just for the audience. It's the non-pharmaceutical intervention index by Olivier Lejeune, right? Exactly. Olivier Lejeune. Thank you. Yes. Yeah, yeah. And MPI stands for non-pharmaceutical intervention. Exactly. I like to call it lockdown. It's much easier. Much easier, indeed.