 Good afternoon. Welcome back. It is my pleasure to chair this session and introduce to you the winner of this year's ESRB research prize in memory of Ieke Bandebu. As the conference program says, the prize is awarded annually to recognize outstanding research conducted by young scholars on topics related to the ESRB mission. The prize was established in year 2014 in memory of Ieke Bandebu, a member of the ESRB ASC, the first ASC, and also a member of parliament in the years in which the financial crisis started. We received a very good number of candidate papers that were assessed at the advisory scientific committee that I have the pleasure to chair. This year, as in previous years, we got a significant number of very good papers on very punctually addressing issues connected to systemic risk and macro-prudential policies. We discussed the shortlisted papers in one of our meetings and apart from the paper that has won the prize, there are other front-runners that will be invited in forthcoming weeks to submit their papers if still unpublished to the ESRB working paper series that the ASC is also editing. The ASC decided to award this year's prize to the paper, and this is a title that you don't have, so this is the surprise of my talk today, to the paper disentangling the effects of a banking crisis, evidence from German banks and counties. There is a paper of which Kylian Hoover, the winner, is a solo author and he's gonna present his paper in a few minutes. Kylian was a PhD student at the London School of Economics, I think, at the time of producing the paper, and was in the job market last fall as a result of which he's been recruited by a University of Chicago. He's currently a research fellow in macroeconomics at the Becker-Fiedmann Institute and later in 2019, he will join as an assistant professor of economics, the Booth School of Business. I don't want to spoil his presentations, any anything about the winning paper. I will make my own remarks about the winning paper after we heard his presentation, but I can say that the quality of the paper is being certified, not only by us, but also by its recent publication in the American economy review. So Kylian, congratulations and the floor is yours. Thank you very much. It's wonderful to be here. Thank you, Javier, to the whole ESRB, in particular Eva and Thomas, who did a great job inviting me and dealing with all my annoying questions, and thanks to all of you for being here today. So this paper is about a banking crisis in Germany. Now, I want to motivate this paper. I'll show you a picture of a very unhappy banker. All his investments are pointing downward, and as we're well aware, there's a huge systemic global banking crisis in 2008 and 2009. Now at the same time, there's a huge global recession, right? Starting in 2008 in most countries, we see here GDP in four advanced economies. I could have plotted this to many other advanced economies. GDP dropped sharply. Subsequently, there was some sense of recovery, but unlike most pre-war, post-war recessions actually, there was no catch up to trend, right? So if you look at the GDP here, it kind of grows on a parallel trend to the pre-crisis growth in many countries without any sign of convergence to the growth path it would have been on had it not been for the recession. So we see a large financial crisis. We see a large recession in many countries, and that's of course not unusual as many of you are aware. You know, banking crises and deep recessions often coincide. I've listed a few famous examples here. Probably the most famous and the most devastating in terms of economic output was the Great Depression in the late 1920s, 1930s, all over the world. We've got the Swedish and Finnish crises in the 1990s, Japan, the East Asian crisis all in the 1990s, and of course the recent recession. Now, an important question is what causes this coincidence? So is it that losses in the banking sector, that reductions in lending in the banking sector cause this type of growth pattern we see? So is there a causal effect running from banking crisis to recessions? That's an important question for a number of reasons. Now, first of all, you know, this whole conference in a sense is about systemic stability, ensuring the financial system works well. And what I'm going to try and do in this paper is to quantify how important, how valuable a stable banking system can be by estimating how large the losses, the real economic losses, are when the economy goes into a banking crisis. Of course, there's also other policy implications of this research. For example, if we know what causes the relationship between the banking sector and the real economy, we might be able to better target you know, the channels, the mechanisms through which banking crisis called perhaps caused real economic losses. Of course, also there's this big question out there in the policy debate, how much should we care about the banks? Is there like intrinsic value to bailing out banks? It's a very contentious issue in the public policy debate. Now, if banking crisis caused real crises, that's a pretty strong argument for trying to stabilize the banking sector once there's a crisis. And then of course, there's still this ongoing debate, which I alluded to just now, which is about why has the real economy not really converged and recovered so slowly after the global crisis? Has this got something to do with the persistent effect of the banking crisis? Are there perhaps just repeated shocks that keep hitting the real economy that have caused the slow recovery since 2008? Now, it's an important question, but it's also a very hard question. And the reason it's a hard question is that this banking system is extremely endogenous to the real economy. In particular, what we worry about when we're trying to estimate real effects of banking crisis is something we technically call reverse causality. Right, so it could be that there's a lot of other stuff going on in the real economy. There's a recession and then banks, you know, they do worse because their debtors default, for example, they get less payback on their loans. And as a result, we always observe banks doing poorly when the real economy is doing poorly. Now, the recent recession is a very good example of this because of course, we saw lots of things going on in the real economy, in particular a housing crisis in many countries. As a result, people who lost a lot of value on their houses had to cut spending, maybe that lecture recession, and maybe that only as an afterthought sort of affected banks. Of course, governments got into trouble all over the world, you know, especially in Europe, but even the US experience of rating down great, there was a lot of uncertainty. So it could be that all these factors actually cause real economic losses. And as a result, these losses cause a banking crisis. Now, what we want to do to overcome this problem is to run an experiment. OK, ideally, we randomly shock a few banks, deplete their capital stocks in various countries across time, and then we see what happens in these economies. Now, fortunately, we're not allowed to do that. Of course, we're not allowed to randomize, you know, huge welfare losses across countries. So we have to find the second best solution. The second best solution is cause the randomization, is what we call a natural experiment. So I'm going to identify natural experiment in this paper and use that experiment to try and cause the identify the effects of the banking crisis on the real economy. In particular, what I'm going to do is I'm going to focus on one large German bank called Commerzbank. Most of you are probably aware that Commerzbank suffered pretty large losses in the recent financial crisis. By the way, whenever I say Commerzbank here, I include Dresdener Bank, which Commerzbank acquired in 2009, so I'm just going to treat them together because both of them, you know, suffered from international investments in the crisis. Now, Commerzbank was in crisis starting in 2007. It was really hit hard in 2008, 2009. By 2011, there's evidence that Commerzbank stabilized again. Now, why am I focusing on Commerzbank? Well, the important aspect about Commerzbank's activities is that it lost a lot of money from international investments, but actually the firms that it was lending to in Germany looked pretty stable, okay? So for Commerzbank, there's strong evidence that firms that were reliant on Commerzbank's loan supply and other financial services would have grown at the same rate as firms dependent on other banks had it not been for the shock to Commerzbank, right? And that's a bit harder to show for other banks, and that's why I focus on Commerzbank here. I should also point out that I'm not on a vendetta against Commerzbank, right? This is a temporary shock against this bank, kind of a mix of unfortunate circumstances contributed to the losses. Commerzbank completely refocuses operations after the crisis. There's actually evidence that it's done much better and stabilized, so no one should go away from this presentation and say, oh, well, that's where you worry about Commerzbank and the firms it's lending to. This is kind of a temporary episode, right? So it's just a picture, a part of the evidence that I have to show you that the crisis was not really caused by what's happening on Commerzbank's real portfolio, but rather on investments in international financial markets. And on this picture, you can see that trading and investment in income essentially explains all of the losses in profit pre-tax of Commerzbank, where it's interest income, which, among other things, includes the income from lending to firms and customers. There's actually pretty stable and increasing and then slightly dropping after 2008. But it's hard to argue that the losses and profits come from the blue line rather than the green line. Here's Commerzbank's lending stock. It drops sharply compared to other banks in 2008, 2009. Interesting perhaps is to see that actually the German banking system is a hold it relatively well. So there are a lot of banks that actually increase lending in the crisis. You can see that in the blue line. It's actually slightly trending upward. Okay, so let me be more precise about what I mean by the natural experiment here. So German firms traditionally form close relationships to banks. They rely on these banks for financial services for a long time. There's a lot of soft information involved in the sense of the banker knows the entrepreneur, knows the guy running the firm and they trust each other and as a result, they start working together. There's also this very German word called Hausbanken, which literally translated means home banks. Okay, you kind of have a bank that you take and you go home. Okay, I'm going to call these relationship banks in English. That's the best translation I could find. So some German firms have relationships to Commerzbank, pre-existing relationships before the crisis. Other firms don't. And the natural experiment is really that some firms are shocked without having caused anything that would make them receive fewer loans such as having lower credit worth. And as some firms are shocked, other firms are not because they're exposed to healthier banks in the crisis and that's the natural experiment. So I'm going to compare, first of all, firms located in the same region, in the same industry, of the same age, similar structure overall. I'm going to compare two firms, one of them borrowing from Commerzbank, the other not. Then I'm going to scale it up, look at the regional level and then I'm going to ask the question, well, what happens if you're not directly exposed to Commerzbank but you're in a region where many other guys around you are exposed to Commerzbank. So are there sort of regional spillover effects even if you're not directly affected? Before I show you the results, let me just give you one more piece of evidence. So there's a firm survey asked by the Ethel Institute in Munich and they asked banks, how do you evaluate the current willingness of your, sorry, ask firms, how do you evaluate the current willingness of banks to grant loans? Are you getting loans pretty easily or is it hard for you to get loans? Now, firms dependent on Commerzbank answer that their loan supply is more restrictive in 2009, 2010 compared to other firms. This is not true before or after the crisis. This is also not true when these firms are asked about product demand, are you struggling to sell products? So it looks like really there's a shock hitting these firms that comes through the banking sector but there's not other shocks in terms of lower demand for their products hitting these firms. And by the way, let me also add here again, the evidence from 2012 onward actually suggests that Commerzbank is expanding loan supply relative to other banks so that firms dependent on Commerzbank report having more generous loan supplies. Again, it looks like Commerzbank managed to turn things around a little bit after the crisis. Okay, three levels of results. I've just mentioned this. First of all, looking at the firm level. So that direct effects when you're directly borrowing from Commerzbank. Second of all, looking at the regional level, a region that's more exposed to Commerzbank on average. Does that region grow more slowly when Commerzbank cuts lending? And then third, what if you're a firm in a high exposure region but with low exposure, you know, directly? So this is the employment of firms dependent on Commerzbank between 2006 and 2012. You see that the firms dependent on Commerzbank grow at the same rate as firms not dependent on Commerzbank. Okay, so the red and the blue line kind of follow similar trends. And then there's the shock. In late 2008, Lehman goes down, the financial system implodes. As a result, these firms are struggling to get adequate financial services. And that's actually real effects. Like I said, they've reduced their employment growth. For about two years, you see the red line growing more slowly than the blue line. And thereafter, the lines are sort of on a parallel trend with not really any sign of convergence, right? Even though these firms now report in the survey that they're able to access bank loans more easily again, there's no evidence that this leads to a real recovery as well in terms of them catching up to the levels of unaffected firms. Think back to what I showed you in the second picture, which was about real GDP, right? The drop in the parallel trend. This is kind of similar. I think that's interesting. The causal effects of the banking crisis seem to resemble what the time series of GDP in different countries say about what happened after the financial crisis. Now, this was just the raw data. I do more sophisticated statistical analyses. I control for the region, the county that you're located in. I control for industry, for size, for firm age, for how dependent are you on exports? How many of your inputs do you get from imports? I compare the results to some customers of other banks that are in crisis, such as the London's Banking, which I've been heavily publicized and so on. But it generally remains true that firms dependent on commerce bank grow more slowly during the years of the crisis, and this is true for employment, as I showed you, but also for investment, and also for patenting. So long-term investment and innovation also seems to go down when you're hit by a temporary banking shock. Second set of results. Now, this is data on how dependent regions are on commerce bank. So what's the average share of bank relationships in a region, in a county or in German Kreis, of firms in that county as aggregated up from the firm level data? You can see the darker regions are more dependent on commerce bank. Quite noteworthy is that the eastern part of Germany is heavily dependent on commerce bank. K-private banks moved into the east relatively quickly after the wall came down. And also some western regions are more dependent on commerce bank, generally for historic reasons. Hamburg, Düsseldorf, Frankfurt, for example, which I've marked out here, they were temporary head offices of commerce bank after World War II, and commerce bank expanded quite strongly around these temporary head offices. So it was broken up by the allies for about 10 years. It had temporary head offices in these three cities. And I used this a little bit of technical speak now. I used a significant instrument for how dependent our regions on commerce bank sort of depends on their distance to want to see historic head offices. So I used both, just the raw difference across regions in commerce bank dependence. I always use this historic IV to get a sense of where does the variation come from in commerce bank dependence. Again, I control for lots of observables that you might think might affect regional growth during and after the crisis. So whether you're located in the east of Germany, what does your industrial structure look like? How dependent are firms in the regions and exports and imports? Is there a London spank that's struggling? How big is the county? How productive is the county? How indebted households in that county? All of that actually makes very little difference. The results generally look like this. If anything, growth of counties dependent on commerce bank is slightly higher up to 2008. And then the growth rate is lower during the years of the lending cut. And subsequently it's sort of a little bit lower, certainly not any higher. And what this implies in terms of the aggregate growth pattern of these counties is that it's very similar to the firm level growth patterns. So there's a decrease during the years of the lending cut and subsequently there's parallel growth again but no evidence of any convergence. Right, so we've established, right, so again, this is just making the point again, you know, there's persistent effects. There's no evidence of convergence of the lending cut. So we've established that there's effects directly on firms borrowing from commerce bank. There's effects on the region overall. And now I want to start the following thought experiment. Imagine you're running a firm. You're not directly exposed to commerce bank. Okay, you're borrowing from other banks. These banks are doing fine in the crisis. But lots of firms around you in your region are borrowing from commerce bank are exposed to the crisis. Now, is that good for you or is that bad for you? Now, economists have written down many models about this, identified many potential channels. And in theory, it's completely ambiguous. So on the one hand, it's kind of good for you and other guys around you are struggling. Why is that? Because they're gonna fire their workers. You probably have to pay these workers a little bit less, right? The regional wage goes down. You can buy your inputs more cheaply. You can buy your workers more cheaply. And as a result, you can grow faster. So great, you just take over the market share of the guys that are struggling to get bank loans, struggling to grow because of that. Now, on the other hand, there might be negative effects. Okay, firms are firing their workers. Now, it could be that as a result, people are buying much less in the region, right? There's decrease in demand. If I'm getting fired, I'm not gonna go to the fancy restaurant. I'm just gonna go to a small chain around the corner. And that's bad for the fancy restaurants. And maybe some firms actually suffer because of a decrease in demand. A third reason, a third channel, a second reason for why firms might actually suffer is that there might be spillover effects. Okay, so it could be that there's some kind of spillover from some guy producing next to me. And there's been a lot written about this in the Silicon Valley, sort of the standard example about all the high-tech firms are running to Silicon Valley. Why? Well, maybe there's something about being where other good guys are that's making you more productive. Okay, it could be, for example, that Apple comes up with a great idea and that's really good for the firm next door because they can steal that idea or profit a little bit from the new technology. It could also be that Apple has a great new idea and now it needs lots of regional suppliers to provide that very particular charger for the iPod, right? And so the producers of that charger are very happy because there's now this input linkage, the input effect that makes firms more productive just by being located in one region. And so that type of spillover effect might be a reason that firms are actually suffering when other firms around them are doing poorly. So what I call these spillover effects, indirect effects, and the headline finding of the paper is firms with no direct relationship, Commerce Bank, but located in high-exposure regions grow more slowly on average. Okay, so it looks like the economic mechanisms are larger in the negative direction. In particular, I can split firms up, I can look at different kinds of firms. Okay, I can look at firms that are producing to sell in the region in particular. They're producing non-tradables, okay? It's hard to sell their products in other regions and so they're relying on local demand. These guys are doing very poorly when other firms in their region are hit by the lending cut. There's also some evidence that if you earn a high innovation sector and other firms around you that are also high innovators are hit by the lending cut, you also do more poorly, you grow more slowly, your employment growth more slowly between 2008 and 2012. Now the mechanism for that could be what I mentioned last, right? It could be some kind of innovation spillover, some input-output channels that are particularly important in high innovation industries that generate a negative effect, a negative indirect effect on other firms in the region. Okay, so this is kind of important when we think about systemic stability because it doesn't look like there's just one channel that leads from banking crisis to firms through a direct financial channel, but these effects spread out through the economy, okay? They filter through the economy through various economic mechanisms. And so far a few studies have looked at firm-level effects, okay, I'm borrowing from a healthy bank. What happens to me? That's gonna underestimate the effects of the regional level because there's these negative indirect spillover effects. So the systemic effects of a crisis at the regional level are larger than what would have been implied by just looking at the firm-level results. So let me just conclude, let me tell you, I had three sets of results. Firms grew more slowly when commerce bank cut lending and they recovered sluggishly after 2010. Regions grew more slowly, counties grew more slowly after the lending cut and they also recovered sluggishly. And in particular, between the direct firm-level and the regional results are these indirect spillovers which are negative. And so they exacerbate the effects of a banking crisis. The effects are worse than if we just look at the firm-level results. I'm thinking about broader picture. What does this mean for policy? Well, a banking crisis can affect lots of firms even if they're not directly exposed to unhealthy banks. This is interesting when we think about a debate that's been going after the Great Recession which says, well, a lot of firms are complaining about low demand, hence the initial shock to the economy must have been a demand shock. The results in my paper, you know, throw some doubt onto this reasoning because even if lots of firms are complaining about low demand, the initial shock could have been a supply site shock, could have been a financial shock. Also interesting when we look at, you know, the growth slowed down in the global economies. Secular stagnation has been a term that's been thrown around. You know, it could be that this is really the persistent effects of a really, really bad banking crisis, perhaps the worst banking crisis we've ever had in the last 100 years. So for policy, you know, if we wanna think about what do we do when the banking crisis happens? Don't just bail out the banks and leave it there. Think about what else you can do. Can we stimulate demand? Can we stimulate innovation to try and get rid of these indirect effects and get rid of these persistent effects? Thank you very much for your attention. Thank you very much, Kilian. I think your very clear presentation was demonstration of the value of the research that we were awarding through this prize. As I said before when presenting you, I didn't want to spoil the contents of the paper by advancing the reasons why we thought this was a winning paper. Let me now elaborate briefly on how the results of the paper qualify for a very good paper and also for a paper relevant to the ESRB. So this paper, as Kilian has shown, documents the effects of weakening of banks' capital position in the case of commerce banks due to, as argued, other assets than loans, other assets than local loans, German loans, but assets possibly localized outside Germany, on the domestic lending of that bank to the firms to which it was connected. Of course there is literature already saying about the effects of a capital shock on the lending that banks get directly involved in. What makes this paper most novel is the fact that it also documents indirect effects on other firms' place located in the same county. The paper shows that local indirect effects are stronger for firms that operate, produce, non-tradable goods, which suggests the importance of local demand, externalities. So this is by commerce bank not lending as much to the firms dependent on it in the counties that are analyzed. These firms are sort of weakening the demand of other firms in the same counties. Therefore, you find these fall in their activity, in their employment, and also you find these local indirect effects to highly innovative firms. The effects are large on impact, and moreover they are persistent as you also document. And the paper offers evidence that the persistency comes from the damage to firms' productivity. You were very modest, you were actually following the directions offering a non-technical presentation, so lots of good empirical results without regression tables. But in the paper, you check that the effects operate through a productivity growth of firms with a high patenting activity in the past in those counties that get affected by the credit crunch. And overall, actually, and I think this is my inside knowledge from Kylian's experience in the job market, this made your job market profile attractive not only to finance people, which is a most standard crowd at least in my own environment, but also to macro people. It's looking beyond finance, and actually, this is why macro proof is so special, exactly in that interface. Methodologically, the paper sustains its results on the state-of-the-art identification strategies. Identification is key. We could find correlations which are totally spurious, and so little on the true causal connection. And you work on the direct effects, but as I said before, on the indirect effects, talking about demand externalities and agglomeration externalities, which makes really your paper have a clear space in the literature. From a macro-prudential perspective, the paper saw several channels through which macro-prudential policy is important for the real economy. So eventually, we are talking about the real economy. Insofar as macro proof can prevent situations in which banks suffer big losses, we will be preventing the real effects that this paper documents. The paper is also important, I think, in the direction of helping calibrate how important is things that we do in the macro-prudential area. So directly or perhaps indirectly, the quantitative results of the paper are saying on the importance of latent parameters for the calibration of macro-prudential policies. So how difficult it is for banks in the short run to replace capital, which is lost due to asset losses, with, say, capital raising the markets, with the consequence being that a shock like the ones suffered during the global financial crisis implied a decline in bank loan supply. You saw clearly this impact. And then you also saw the lack of substitutability of the supply of credit provided by commerce bank, at least for the firms whose employment suffers due to it, which is another channel we take into account when thinking that credit cycles might be bad for social well-being. It's not just that credit by banks can be replaced easily by other credit, at least not in the context of the recent crisis. And finally, you illustrate about externality. So the effects are not only for the affected firms, the effects are also for other firms in the same counties. You don't have data or you haven't explored the household angle, but I guess if commerce bank was also involved in lending to German households, you will have similar demand effects operating possibly through household credit. So, okay, with this in mind, we have a few minutes for questions from the floor. If you have, I guess Kilian will be pleased to answer or elaborate. If not, let me close the session. We have a spanded coffee break by formally awarding the price. It's a virtual price so far to Kilian. Please join me in applauding him for his presentation. Congratulations.