 Good afternoon to the last round of this two days conference. Welcome to all of you in the room and welcome, again, to all of you out there in the world on Webex. And thanks for holding on. So we have a very exciting last session with two papers. And we start with our first paper on domestic climate policy and cross-border lending. And the presenter is Emanuella Beninkasa. Emanuella, the floor is yours. OK. So thank you. Thanks a lot for having our paper on the program. And thanks for all of you that are still here in the room. So today, I'll present this project, which is joint work with Ghazi Kabash and Steven Nogina. And I'll start by motivating the paper. So climate change is a global challenge whose solution requires global coordination and cooperation. And this is something that has been recognized also by policymakers and international institutions. However, what we observe still nowadays is that there is a significant heterogeneity across countries in terms of stringency of their climate policy. And what you can see in this slide is a world map reporting countries worldwide depicted in different colors, with green colors showing countries which are more climate active, so they have a better climate performance and therefore a higher climate policy stringency compared to countries which instead are depicted in darker colors, so orange or red, which instead have a lux climate policy stringency. And if we zoom in in Europe, we can see that this heterogeneity is there among European countries. And also, if we compare European countries to countries worldwide, for example, Canada, Russia, or Australia. So there is nowadays this heterogeneity in terms of climate policy stringency. Now, what this implies in terms of when we look at the domestic market, when we look at the domestic market, stringent climate policy can have two different type of implications. The first one is that a more stringent climate policy can increase the demand for funds to invest in green technologies or innovation from firms. However, what we know from the academic literature is that bank finance is not well suited for this purpose. Why? Because banks are reluctant when it comes to investing in risky assets or because the underlying collateral may lose value. And the second implication is that a more stringent climate policy can require a change in a firm's business model and also their production process. So this can decrease firms' profitability and also possibly because of this reason make domestic lending less appealing. If this is what we know from the academic literature, little is known, however, about the fact that bank lending, so bank finance, has a cross-border in terms of stringency of climate policy. So whether a more stringent climate policy has implication for bank lending across borders. Specifically, we do banks react to our more stringent climate policy. And if so, our banks are focusing their lending to brown firms located in brown countries. So what we do in our paper is try to answer this question. And specifically, we provide evidence that banks exploit the lack of global coordination climate policy by increasing cross-border lending to brown firms located in brown countries. And we do so by using a global measure of climate policy stringency. So we compare countries global in terms of the stringency of their climate policy by using the CCPI index. And I will tell you more about this index and estimate the fact of climate policy stringency on cross-border lending in the syndicated non-market. Because of the nature of a syndicated loan, we are going to isolate the credit supply and control for credit demand drivers by using a loan fixed effect in our estimation. So this means that we are going to compare the credit supply of banks joining the same syndicate and providing funds to a foreign firms located abroad. Importantly, we also try to control for omitted variables. And in addition to using a loan fixed effect, we use a change in green party shares in parliament to instrument the CCPI and to estimate the causal effect of climate policy stringency on cross-border lending. Now, let me preview the results quickly. Let's take the case of two banks joining the same syndicate, a bank located in the United States and a bank located in Germany. And they both join the same syndicate, providing funds to a firm located abroad. Now, let's take the case of year 2015, where Germany had a more stringent climate policy according to our index by six index points compared to the United States. What we find is that the German bank would increase the cross-border loan share to the firm located abroad by 0.5 percentage points, which corresponds to 6% increase relative to the mean. And this is, of course, within loans so compared to the American bank. At the same time, we also find evidence that the same German bank increased the cross-border loan share to a firm which can be defined as a polluting firm, so a high carbon intensity firm by 5.5%. However, also decreasing the domestic lending to a brown firm located domestically by 15%. So this suggests that a bank is indeed engaging in this refocusing of lending to brown firms located in brown countries. So quickly on the contribution of our paper. So our paper contribution is threefold. First of all, we contribute by showing that cross-border lending is a tool to protect loans' portfolio exposure to transition risks. And there is already evidence in the literature showing that firms engage in regulatory arbitrage if they are located in countries with higher, with more strict climate policy. And they do so by shifting their operations output and production to plans which are located instead in areas with less climate policy. Second contribution of our paper is to study the role of banks in promoting green financing and help the transition to a low-carbon economy. And there are studies already looking at fossil fuel lending and also substitution between bank and bond financing. And finally, we also show that climate policy stringency can be an incentive for cross-border lending. And the academic literature has already studied the determinants for cross-border lending by looking at geographical proximity, cultural sharing aspects between borrowers and lenders, and also by looking at regulatory arbitrage opportunities. So I'll jump to the data and identification. So how do we measure climate policy stringency? Measuring climate policy stringency globally is not an easy task. Why? Because each country can have their own measure and also their own objective and also output coming out of the policy. We do so by using a global index, the Climate Change Performance Index, or CCPI. And this index is constructed by a non-governmental agency called German Watch, and which is located in Germany. The index is a country-year level index and covers 57 countries worldwide. The index is composed by four main components, greenhouse gas emission, which accounts for 60% of the index, renewable energy 10%, energy efficiency 10%, and climate policy 20%. So as you can see, this index it's complete in the sense that it tells us about the input in terms of policy, which is indeed provided by this component, but also in terms of output. So what is really the performance of each country in terms of outcome coming from more stringent climate policy? Now, you may question, why are you using an index to measure climate policy stringency at the country level? We argue that there are many advantages. The first one is that an index makes global comparison possible and easy. And also, there are many different type of policies worldwide or measures by government. So the CCPI allow us to have a comprehensive measure worldwide, which allow us also to compare climate policy stringency across countries. Importantly, the CCPI is also recognized by international institutions and policymakers as a measure to indeed assess climate policy stringency at the country level. For example, it is presented at the United Nations annual conference on climate policy and also used by policy institutions like World Bank, FSB, and also practitioners in the industry. So in this slide, you see the variation in the climate policy in the CCPI index. And you can see that European countries, on average, which is depict on the x-axis, have a higher score compared to other countries, for example, Australia, Canada, or Saudi Arabia. But at the same time, you also see that the index, there is quite a lot variation within the index, which is indeed depicted on the y-axis, showing the standard deviation of the measure itself. How do we measure cross-border lending? We use syndicated loans from Dilsken. And our sample comprises only observable cross-border loan shares in the period 2007-2017. And also loans, the way we define a cross-border loan is by looking at loans provided by a bank to a borrower with different nationality. In the firm's location, we are going to look at the headquarter of the firm. In terms of bank's location, we look at the country where the bank is located. And finally, we also hand-match the Dilsken data with bank scopes to know more about bank's balance sheet. In terms of identification, as I explained before, during the preview of the results, we are going to regress the cross-border loan share on the CCPI. And we face two type of challenges. The first one is that, of course, we want to identify credit supply facts and control for the mind drivers. We do so by saturating the model with the loan fixed effect, which allow us to compare banks joining the same syndicate and control for loan characteristics and also borrower characteristics. This is very granular and also clean in terms of fixed effect. Secondly, there can be still concern in terms of omitted variables. We run two different type of exercises. The first one is that we control four variables which are associated both with the climate policy stringency and with cross-border lending. And finally, we instrument our CCPI with Green Party sharing parliaments. And I'll show you results on the relevance condition and also we have a few exercises in order to argue on the exclusion restriction of this instrument. So I'll quickly go to the results. In this slide, you see the baseline results from our specification, where we regress the cross-border loan share on the climate policy stringency of the bank country. And at stages, we control for bank controls, borrower fixed effect, ear fixed effect. In the third column, you see results when we, sorry, the fourth column, where we have borrower times ear fixed effect, alakwaja imian. And finally, column five, you see the results where we saturate the model with the loan fixed effect. Now, I want to spend a few words on the results from column five. And so what you can see is that the coefficient is positive and statistically significant, telling us that a more stringent climate policy increased the cross-border loan share. Now, in terms of economic magnitude, what this coefficient tells us, so for a German bank, which is indeed located in a country with high climate policy stringency, these 0.04 percentage points will tell us that the German bank increased the cross-border loan share by 6% on average, compared to an American bank instead located in a country with lower climate policy stringency. And we also run an additional exercise where we control for bank fixed effect and the bank times ear fixed effect. But the results are still there. In terms of omitted variables, we have a table where we try to see the sensitivity of our results, where we, at stages, saturate the model with the bank controls, economic controls, culture control, so control for the culture shared by the borrower and the bank. We also control for demographic aspects, bank regulation, and institutional quality of the country where the bank is located. But throughout all specifications, the results are still confirmed. In terms of IV, we instrument the CCPI with the change in Green Party share in the national government. And the first column shows the results from our first stage. The coefficient is statistically significant and positive sign. And also the F start allow us to be confident when it comes to the relevance condition, so being above the threshold. Column 2 to 4 shows the result from the second stage where we regress the cross-border loan share on our instrumented CCPI. And specifically, let's focus on column 4, which is the one showing the results where we saturate the model with the country controls, bank controls, and also loan fixed effect. The coefficient is statistically significant and positive, showing that we can interpret our results as causal outcomes. We have a higher magnitude, but possibly because of measurement errors. We also have a few exercises where we try different type of instrumental variable where we look at the, we argue on the exclusion restriction of our instrument. But for the sake of time, I'm going to skip this. And you can refer to the paper for more details. In terms of mechanism, so the results so far show that a more stringent climate policy leads to an increase in cross-border lending. But why is this the case? And the question is, what is the mechanism at a play? Our conjecture is that there is a raise to the bottom in the sense that the heterogeneity in climate policy stringency across country can be viewed as a form of regulatory arbitrage. And banks can circumvent climate policy stringency, and especially the cost that this brings on their loan portfolio through international banking. And our conjecture is also that banks may want to increase their cross-border lending to protect their loan portfolio and to lower the cost of climate policy stringency, which of course will lead to a raise to the bottom behavior. We try to show evidence for this type of mechanism through different type of exercises. The first one is where we look at the climate policy stringency of the country where the borrower is located. And we do so in two different type of exercises where we interact the CCPI of the lender country with the CCPI of the borrower country. And you can see that a more stringent climate policy of the country where the borrower is located decrease the cross-border lending for a bank that is indeed located in a strict climate policy environment. And the results are also confirmed where we split the sample differentiating between countries where the bank is located in a more stringent climate policy environment than the one where the borrower is located. A second exercise is where we look at what's happening in the domestic market. And we do so by having a model where we have a triple interaction. Specifically, we are going to look at firms which are defined as polluting firms, so firms with a high carbon intensity risk. And when we look also at the case of a loan which is granted to a firm located in the domestic market. So this is identified with the same country dummy variable. And let's focus on the results. On column five, you can see that for a bank located in a country with high, with stringent climate policy that lend to a domestic firm, which is defined as a brown firm, so a polluting firm, the share, the loan decreases. However, and interestingly, this is not the case where instead we look at polluting firms which are instead located abroad, where we find a positive coefficient in our estimation. Meaning that these banks increase their lending to foreign firms, which are polluting firms and located abroad. Now, we have more exercise in the paper, but the follow-up question is why is this happening? So we question whether this is happening because of more stringent climate policy increase the cost for firms and therefore lower corporate profits. And we run an exercise where we look at the corporate profits at the country level and the correlation of those with the CCPI. And we find a negative correlation, meaning that higher climate policy stringency lowers the corporate profits of firms located in stringent environments. And if this is the case, we also question whether higher climate policy stringency decrease the corporate profit for firms and therefore this also has an impact on the portfolio of banks. And we look at the non-performing loan ratio of loan portfolio and also net profit. And we find that a more stringent climate policy is positively correlated with NLP and negatively correlated with profit ratio. However, for banks that are defined as cross-border banks, this is not the case in the sense that these banks are going to lend abroad and therefore decrease the NLP ratio and also increase the net profit ratio. So this is showing that the banks trying to circumvent the costs of a more stringent climate policy. Now, let me conclude. What we try to do in our paper is to investigate whether banks use cross-border lending as a tool to react to a more stringent climate policy, which brings up higher costs for banks and also lower corporate profits. We find evidence that banks exploit uncoordinated national climate policy by refocusing syndicated lending from green to brown firms in brown countries. And a tentative policy takeaway from our study is that a lack of policy harmonization may trigger risk to the bottom behavior by banks and also threaten the effectiveness of climate policy by countries. Thanks a lot, and looking forward to the discussion. Thank you, Emanuella. Discussant of the paper is Glenn Shippens from ECB's research department. Glenn, please. OK, so good afternoon, everyone. Thanks for inviting me to discuss this very interesting paper. So I think Emanuella did a great job in conveying the message that this paper is really all about the fact that a lack of cross-country coordination in terms of climate policies can lead to some sort of regulatory arbitrage by banks. So what they're going to show in the paper is that lenders tend to increase their cross-border syndicated loan shares when the climate policy stringency, when climate policy at home is more stringent. I think this is nicely summarized in this picture that comes from the paper. So here on the x-axis, you have the difference in climate policy stringency between the home and the foreign country. On the y-axis, you have the lender share and you see once this difference becomes positive, so once climate policy stringency is higher in the home country than in the foreign country, these cross-border lender shares go up and become larger. So overall, I think this is very much reminiscent of another, say, banking, a very closely related banking literature on the bank's reaction to other types of regulations. So for example, there's work out there showing that if capital regulation is more stringent in a home country, banks tend to move some of their activities towards countries where it's less stringent. So it's a nice addition to this literature. And overall, I think it's a very interesting paper, well-executed empirical analysis, very interesting read, and has some very important, potentially findings with very important policy implications. So the downside of this is that it makes my job, as discussed, a bit harder because it's hard to find some very good points here. But what I'll do is try to mainly focus on the interpretation on some of the findings in this short discussion. So I'm going to make three comments. So a first question I want to ask is how certain can we be that what you capture in the data here is really about banks trying to avoid climate regulation and not potentially about something else? And one thing that I'm a bit worried about is that the main results could be partly driven by banks trying to avoid a physical climate risk instead of actual climate regulation. And in order for this to be the case, there's basically two assumptions that need to be fulfilled. So one, physical climate risk would need to be correlated with climate regulation. And two, physical climate risk would need to be correlated with cross-border lending. And if that's the case, you're running to a very standard type of omitted variable bias that the analysis might suffer from. So on the first point, I quickly downloaded the indicator that you're using. So the CCPI index is on the y-axis here. And on the x-axis, I have another index from the same data provider that captures physical climate risk in the country. So what is very simple, this very simple figure shows that at least there's some correlation typically between physical climate risk in the country and actual climate regulation in the country, which makes sense. I mean, I think government is more likely to take action when they're facing larger climate risks. So this first assumption, there seems to be potentially some truth in that. So for the second assumption, the link between climate risk and cross-border lending, I don't have any data to back this up, though. I would like to say that this is somewhat consistent with the actual analysis in the paper, so that low capital and high MPL banks seem to be more likely to react by having more cross-border lending going on. This is because, if you think about physical climate risk, the way that this would affect banks is exactly true, higher credit risk, lower profitability for the banks, and hence, potentially an incentive to move to cross-border lending. Now, I know that you're doing a lot in the paper in terms of trying to avoid this type of omitted variable buys, but I'm not sure that the IV setup at the moment solves this exact issue because there is quite some work out there that shows that the exposure to climate shocks also influences green concerns of the population in the country and also affects, for example, the voting behavior. So if this would be the case, then there's a bit of an issue with the exclusion restriction for the IV setup here. Now, it's not all bad news, this first comment, because it's actually very simple to address with a number of robustness checks. You could consider a robustness check, for example, that excludes home countries with high physical risk. You could try to control for this high physical climate risk in your analysis and so on. So there's a lot of things you can do to cover for this, but I think it's something that's worth checking. So a second comment I have is on the mechanism in the paper. So why does climate regulation leads to more cross-border lending? And the main mechanism that the authors have in mind is a very straightforward one. So climate regulation is basically expensive for firms on a very general level. Firms need to do green investments. They might be confronted with stranded assets. They might be confronted with carapaceous and so on. So in the end, this will lead to higher credit risk in the portfolio of the bank and potentially lower bank profits, giving banks an incentive to move their lending abroad. So as a short aside here, so there's lots of potential sub-mechanisms here. I think one of the interesting things to do could be to zoom in on some of these sub-mechanisms, such which is, for example, done in a paper of two by two colleagues of mine that very much focuses on differences in carbon taxes across countries and how this affects cross-border lending. That's more as an aside. So what I really want to focus on here is a potential alternative explanation, alternative story that I think is also very consistent with most of the results in the paper, but has a bit more of a benign view towards the banking sector if you want. So I think a very simple alternative explanation here is that banks might simply be following clients abroad. So there is a literature out there by now that shows that, and you also refer to it during your presentation that firms react to this climate regulation by also potentially moving production abroad. So it could very well be that the bank is simply following its clients abroad by, for example, starting to lend to subsidiaries of these domestic firms in other and foreign countries then. And importantly, I think this is consistent with a lot of the findings in the paper. So for example, one of the things you find is that banks keep, when they lend abroad, they're going to do that in sectors in which they're also specialized at home. But of course, if they would be lending to the same firms, this would logically follow from that as well. So I think it's worth checking to what extent the results might be driven by lending to foreign subsidiaries of these domestic firms. And again, this is something that's fairly easy to check. That's the good news again here. And then a last minor comment is on this CCPI index. So to what extent is this CCPI index a good proxy for climate policy? Because as you mentioned during the presentation and as you're very open about in the paper as well, climate policy in the end is only makes up 20% of this actual index. So a lot of the things in the index capture other things related to climate and being green, such as energy use in the country, the percentage of renewable energy available, the number of greenhouse gas emissions in the country, and so on. So for me, this index mainly captures how green an economy is. And how green an economy is not necessarily 100% driven by regulation. So things like the sexual composition of an economy, the natural resources that are available, and so on, will have a big impact here. So it was a bit puzzled why you not simply use this 20% climate policy component of this index, because you do have the data on that. And you do use it in the robustness table in the back of the paper. And the robustness table shows that your results basically hold and they stay very strong when using this more specific climate policy index. So why not use it for the main analysis? It's a bit of a puzzle to me. And I think it would make more sense. It would make the results that you have even stronger. So apart from that, the rest is just some very minor comments which we can discuss later on. So overall, I think a very interesting paper definitely worked your time reading very well written. Already a very extensive analysis. And I'm best of luck with the paper. Great. Thanks, Glenn. So I opened the floor and also on Webex and in the chat. Anyone who's there then? That's a very quick question. Do you exclude the leader ranger? Because the concern could be the potential borrower seeks a bank that is under a brown potential borrower seeks a bank that is under a loser regulation from a country with a less strict climate policy that would suggest to remove the leader ranger from the sample of banks. Of course, there are others who are just participants with them. So you have a loser relationship, if any, between the non-leading participant and the borrower. I found the results very interesting. I just had a question about how to interpret them or what are the policy implications. Like Glenn said, it's quite similar to cross-border lending being used to bypass capital regulation, for instance. But for capital regulation, the point is to make the bank safer. So it's really a problem that the bank is taking the same risk in a different country. For environmental regulation, we are not really targeting the banks themselves who are targeting the firms. So showing that the bank is no longer lending to polluting firms in a given country, lending somewhere else instead, I think we care about this only if this means that the firm that receives credit would not have received credit otherwise. So you see what I mean that I think the policy target is at the firm level and not at the bank level. And so that's a big difference, I think, with the literature on capital regulation and cross-border lending. Thank you. Thank you, Daniel Guch, ECB, SSM. Very, maybe marginal, but to me, interesting methodological question. I read your paper, which is very interesting. And I focused on the point where you use shares of a green party in the parliament as a kind of control variable for stringency of environmental policy. Maybe I'm simplifying it too much, so I leave it to you to explain exactly what your purpose was, but just to observe. Well, green parties, when they get into parliaments and get stronger in the parliaments and again in the government and tend to lose their original radical approaches, this is the case here in Germany as in many other countries, they not necessarily represent as stronger as most stringent policy, they make more compromises, they are represented in the power. So I would be interested in this particular methodological aspect, how you use it and how you take account of various political situations that are imaginable here. Any other comments, observations, questions? If this is not the case, then Emmanuel and Laura Sios again. Yeah. So thanks a lot, everyone, for the questions and thanks, Glenn, for the discussion. I'll start by commenting, trying to answer a few points in the discussion, and then I'll go to the questions. So I like very much the physical climate risk correlation that you show with the CCPI, because we didn't think about that, to be honest. And I think it's a nice point. And I'll wait for your slides to know more about this score. However, you also mentioned this during your presentation, so maybe what we can do, while you were talking, I also thought about maybe a possible exercise can be tool at different countries, like countries that are more exposed to extreme weather events and not extreme weather events. But I think the CCPI itself, it's an index which accounts for climate policy in general, also in terms of output. But I understand the point, and maybe we can look at this, or also control for possible climate awareness in our regressions. And about the point on banks following firms which are located abroad, I mean, this is, I'm not sure, but this would imply that the results would be demand-driven and not supply-driven, because then this should come from the demand of credit by the firm if I understood the point. And I mean, we try to do our best to be a loan-fixed effect, but I'm happy to pick up again, because maybe I missed something. And finally on the proxy, if the CCPI is a good proxy for climate policy stringency, yes, we also tried with the component of climate policy. And the idea for us was not to use the component itself, because then we will not account for the output of the policy itself, which is instead measured via the greenhouse gas emission. So if you have a stringent climate policy, which is measured by the component of 20%, you should also observe some results on the greenhouse gas emission. So this was the idea for us to have a more comprehensive measure. But we have tried with the climate policy component, and the results are there. But again, thanks a lot. On the questions, so the first question was about whether we excluded the leader ranger from our regression. So we have a table where we look at the split between leader ranger share and participants. I did not have the time to show this, but we do not find a difference in the results, meaning that it's not because of the role of the bank in the syndicate. And I mean, this is kind of hopefully helping our story. But yeah, thanks a lot. And the second question was about whether, yes, the second question was about the interpretation of the results and whether we can really interpret the CCPI, if understood, as a policy which is affecting, I mean, it's not really affecting the banks, but rather the firms. I agree with this point, but indeed, that's why we do many exercises in our paper to try to show that actually a more stringent climate policy brings up higher costs for firms. And therefore, that's why banks are reacting in the following way. Again, I agree with this interpretation, but for us, the objective was to compare, have an analysis which is at the global level. So that's why we also don't look at a single type of climate policy measure, and that's why we have an index there. And about the last point, I guess you are referring to the table where we regress the Green Party share on country level variable, macro level variable. The idea there is to see whether macro variables like GDP per capita, GDP growth, are actually correlated with the Green Party share, because we don't want that. Otherwise, that would be against the exclusion restriction. And the idea there is simply to see whether there is correlation among the Green Party share and the macro level variables and vice versa. So whether macro variables are determined by the Green Party share. And the idea is just to argue on the exclusion restriction. But I have to discuss afterwards on this myth. Thanks a lot, everyone.