 I think we have to go now and listen to the stress test perspective and discussing their effectiveness as well. So we end from Haleu Institute for Economic Research, you have the floor. So thank you so much for having me here to present this work in progress paper. So today I will talk about climate stress, bank lending and the transition to the low carbon economy. This is joined with Larissa Fuchs from University of Cologne, Claude Sheik from University of Bristol who is also in the audience and Chang Wen also from University of Bristol. So why do we care about this topic? Recently we have a lot of debate on essentially what would be the role of banks in financing a green transition. So in the one hand there has been call for more disclosure on the carbon emissions linked to the lending activities. On the other hand there's a lot of skeptical views for example what if banks need to also stick to the polluter so that they can support the green transition. On the other hand for example Gleason at all 2022, they raise a question that financing a climate change risk at the moment without a clear expectation from regulator would be quite unrealistic for banks to change. So let me give you a little bit of other debate from the literature as well. So in the literature we learn a little bit about banks account for climate risk in their supply chain, their credit supply and also risk management. So for example back up is to get all 2021 and even off at all 2022. But there the results really mix because not all banks are the same. Some of them may not be well prepared for the green transition because they have also private interest to protect brown borrowers. Think of for example legal see effects so they would have relationship lending with brown borrowers that are really profitable. So in that case this is not so clear whether they would really change the financing strategy to take into account the generation risk. They can also ship the generation risk away for example think of selling off to the less regulated entities. And they can also lend to brown borrowers abroad which we will learn a lot later today by Benesca at all 2022. And the danger of financing green transition with our clear expectation is that it could actually create what we call greenwashing behavior that means that banks finance to appear more green and that could lead to for example green bubbles. And later on we will also see that there could be financing green transition without clear expectation lead to the disconnection between sustainability reporting of the banks and also lending activities by Maria Suntow later on. So against this background we ask a question that can bank supervisors somehow influence bank decision to support the green transition in a more sustainable way. So if we look at this figure here so the figure on the left hand side you see this is the number of central banks around the world incorporated climate change in their policies objectives. So the blue line show you that around 50 central banks have participated in green network membership like NGFS and the purple line show you that there are around 10 central banks as of 2019 have green lending guidelines or at least a green bond program. So our research question is that I would like to bring you from Frankfurt and then we go to Paris and we look at the French climate threat test of 2020 as a natural experiment and we combine this practice with a measurement for from environmental risk from red risk and we ask two questions. The first question is that how do climate threat tests affect greater supply and the cost of credit for firms with high exposure to the recent risk and later on it will work you through the result that we show yes it does have an effect possible without climate threat test banks seem to build this generation risk from firms as a short term perspective that means they reduce credit but they did not price the risk into loan contracts with the climate threat test however because it serves as an information collection net that means that threat test banks are better informed about the exposure to generation risk in the long run. So instead of reducing the loan volume to those high generation risk firms they actually increase the loan volume for those browner firms but at the same time they charge higher interest rates especially in case that they have lending relationship with those brown borrowers. The second question we ask is that so we see some movement in the lending activities how does this translate into the final activities of firms meaning that can firms transform to be better whether they pick into account environmental risk when they invest in their projects and also whether they make any effort to reduce capital emission and green their supply change. So here I just would like to give you some institutional background on what exactly has been done in the French climate threat test. So in this threat test the central bank do not view transition risk as something like they would like to kind of support just the green transition but they look at it from the financial stability point of view. This means that let's imagine a little bit unimaginable. Let's say in 30 years what would happen when carbon prices increase sharply and the bank balance sheet exposed to a large losses. So by this exercise it's a little bit of the alignment between the incentive of central banks and also with the incentive of banks and borrowers. So from 2020 until April 2021 the French supervision and resolution authority was the first to run this climate threat test and they invited banking groups and also financial and also insurance companies to participate and estimate the losses that could happen when we have severe scenarios in the economy. For the insurance company they focus more on physical risks whereas for the banking groups they focus more on the transition risk also because of some data availability. So in this test they have three different scenarios that you see in the graph here. So the baseline is the black line there where we have an orderly transition. This means carbon prices would increase gradually and would reach around 200 euro per tons CO2 by 2050. Then we have a delayed transition which is in the yellow line. There we see that nothing happened until 2030 so regulators do not act and there's also little change in the technologies or firms that can transits into a greener activities. In the red line we have an accelerated transition. This means that we take a really aggressive action from now. So around 2025 the carbon prices are increasing and it's really sharply increasing and reach around 900 euro by per tons CO2 by end of 2050. So in this exercise nine banking groups in France actually chose to participate in this test and they estimate the expected loss from these scenarios and these nine banking groups represent 85% of French bank total assets. So in this scenario we see that this is quite a long-term and forward-looking exercise. From 2021 until 2025 we use just static balance sheet and then based on the assets portfolio of banks at the time of the climate threat that happened and then estimates the loss whereas after 2025 they take a dynamic balance sheet approach where banks can actually relocate a little bit in terms of lending to different brown versus greener firms. So in this practice as Michelle I actually talked with me yesterday this is more than like a learning exercise so they do not identify any kind of violators but this actually provides a platform as a two-way feedback between supervisors and the banks so they produce new information because it's not just some victim about maybe Jean-Lucen Riggs would affect the bank balance sheet but it's put a number. This means that it could be expected loss if we do not act now and if the carbon price is pushed to be higher in the future then the bank need to care about how the loss would affect them. So the data we use that we use syndicated lending data for all French borrowers between 2017 until 2022 from Dealscan and the loans to these French borrowers can be provided by both French and other non-French banks. We remove all the financial firms there and we merge the data bank scope to obtain lender characteristics and then we obtained borrower characteristics from Compute Start and Amadeus. The reference data give us an environmental risk index we call ERI that based on 10 100 different sources actually 100,000 different sources for example from newspapers also like third-party reports and reports from the firms and even social media that talk about the bad incident happened to the firms related to for example pollution and carbon emissions. So previous paper for example Duane et al already shows that this risk index could imply the level of chance risk of the firms and have a little bit more like unbiased measurement because this is not just based on how much the firm report to the public whether they are exposed to the chance risk but it has like third-party source of news as well. So here because the nature of this climate threat that is that French bank could choose to participate so we may actually suffer from the selection bias so to fix this we then form a control group banks that are non-French banks who cannot select to participate in this threat test and they are similar in the sense that they also provide loans to the French borrowers and let me give you a little bit more information on why this ERI environmental risk index could capture chance risk. So in this graph you see the ERI for seven different industries so we aggregate this ERI from firm level to the industry level and we see here that the higher ERI imply higher chance risk so let's see mining oil and gas here have the ERI the highest in the sample of 27 and then we have manufacturing utilities transportations being their second highest. So in this measurement you see that it's not just a correlation with the level of government emission in the industry but it also has a little bit of forward looking future because for example mining oil and gas are the industries that harder to chance this meaning that there's not much of the technology that could give these activities being more green whereas we have manufacturing utility transportation there are still hope in this sector because we could actually have green technologies and have more renewable resources that can be used in this sector summary statistic we have in our sample 1758 loans given to 81 French firms by 126 EU banks so we talk about really large corporation and here once indicated loans has the average side of 400 million euro the maturity of four years loans press a 50 basis point 42 percent of the loans given by the stratested banks and 58 percent of the loans given by non stratested and really similar EU banks in our sample 30 percent at the firm level we see the 30 percent of firms getting loans from stratested banks after the climate stratest they have the ERI ranging between 0 and 27 the carbon emission growth ranging between minus 47 up to 112 percent so identification strategy will be really quick here we in the first step when we look at the landing we do two steps first of all we just look at the relationship between loan volumes and credit spread as a function of the firm ERI so essentially our coefficient of interest is going to be taught here so this is going to tell you whether higher ERI firms mean higher Chandrithan Riks firm would receive more or less credit and whether they have any kind of premium taking into account the Chandrithan Riks that they face in all regression we have firm fixed effects to control for demand side we have bank time fixed effects to capture bank specific time varying effects we have loan type fixed effects which distinguish between revolving and term loans we have loan purpose fixed effects which is about whether the loan is for investment purpose or buy it out and this is really strong control for the demand for specific type loans in the market in the second step we do the interaction with climate stratest so we have a triple interaction here we interact ERI of the firm with both T which take value of one after the climate stratest in quarter two 2022 the treated B going to be one for French banks that participated in the stratest and zero for non-French bank who cannot choose to participate but also similar so in order to tease out the effect of really climate stratest rather than the change in the Chandrithan Riks firms here we fix the ERI to the preshock level and then the second step then we just link this to the firm level environmental performance and we run similar regression but at the firm level where we have the triple interaction ERIF interacted with whether that firm receive any loans from the stratested bank after the time of the stratest T and here we look at the dependent variable going to be real environmental performance of firms in terms of short term adjustment and also long term adjustment for short term adjustment we have like COO reduction target and whether firms declare that they have environmental training for employees whether they have any initiative to restore the environment and whether they include environmental project evaluation in their investment for the longer term adjustment we look at carbon emission growth both total and scope one emission score whether they terminate relationship with environmentally unfriendly suppliers and whether they source materials based on environmental criteria so in order to convince you that we really capture the effect of climate stratest we need to know that stratested banks in France are similar to non-stratested banks in other EU countries so we first of all look at the similarity in terms of firm environmental performance that link to the two type groups and we see that prior to the stratest borrow or receiving loans from stratested banks and non-stratested banks have similar ERI changes so here we have the mean of the treated group control group and then the last column it shows you the normal night difference in difference this is actually show you that if this normal light difference is larger than 0.25 then we should worry about the non-parallel trend assumption and here we see that in terms of also bank characteristic and firm characteristic they are quite similar prior to the climate stratest so let me then give you finally the result the first set of results here just show you the relationship between bank lending and borrow ERI and we see that the first two columns show you the effect on loan amount and the last two columns show you the effect on spread so when the firms ERI increased by one unit here you see the loan amount that the bank given to them before the climate stratest actually declined significantly and equivalent to between 7 to 13 percent whereas for the spread we do not see any kind of adjustment that banks price the generation risk in loan contracts so now we look at the effect of climate stratest we should look at the coefficient on the triple interaction there and in the first two columns we also have the effect on loan amount the last two columns we have effect on spread and we see that following climate stratest loan volumes increased significantly for borrower with greater generation risk and loan spread also increased significantly so it seems that we have some expectation maybe climate stratest at bank seem to aid borrower to the generation toward greener activities but they adjust the risk accordingly and they ask for additional night by this point so here we just do a little bit of split sample and we see that our results actually mainly driven by their relationship lending this means that banks give more credit to the firms that they have long-term relationship with and they also adjust more loan spread for these firms but whether these changes in bank lending could really lead to changes in firm environmental profile we look at all the short-term adjustment here and we see that climate stratest banks after they give loans to these firms it seems that their firms have like more likely to have objective to save resources in their activities they also more likely to have environmental training for the employees especially they committed more in the emission reduction targets but let us look at the long-term adjustment here we see very little adjustment the only things that we see here is actually a small increase in emission score but we have not yet seen any change in total emission growth or direct emission growth we also do not see any changes in EHG score or changes in their domination with environmentally unfriendly suppliers so we see here this is that it could be two ways first of all we could worry about whether this essentially have real effect to the long-term transition to the green economy but the other way that we could be a little bit more positive is that it could be these things need longer time to adjust so this is why we are hoping to continue this project and see whether we could see really real impact in the long run in possibly one or two years so this is still a work in progress so I have listed new next step here and I do not go through all of them I just wanted to mention that we would also like to do a little bit of iso no validity because we know that the ECB also does climate stress so we will replicate our result with essentially the EU climate stress that done in 2021 so let me conclude we believe that we share some kind of new information on how bank climate stress directly affects their bank lending and also firms environmental performance so we see that there's a kind of still mixed result we see banks do change lending activities but in terms of firm environmental performance we see some changes in their kind of more easy to adjust this means that firms declare more that they would become more environmentally friendly but the hard information on whether there's a real adjustment to a lower carbon economy is still something still missing there so this is to be continued thank you so now I would look forward to the discussion who will join us online thank you so we have now Tristan Jouard from Bonne de France this is very French session by the way Andreas you have a French chair two French discussions and we talk about the French stress test anyway thanks for this okay more seriously over to Tristan now for another thank you hello everybody thank you Andreas for the invitation to discuss this paper and thank you Yann for the very clear discussion so overall we really enjoyed reading this paper and I think it make a significant contribution to the debate on the effect of climate regulation on financial institutions sorry so more specifically it it look at the effect of the climate stress test on bank lending and the environmental profile of the borrowers it uses a casual experiment based on the French climate stress test and employ a triple a definitive approach to estimate this effect so the main results are that the treated banks tend to increase their loans to the polluting companies which is not very intuitive and they also increase the interest rates on the same loans and also the paper shows that the polluting borrowers tend to improve their environmental profile at least in the short run so the paper has many qualities first it is on the timely and relevant topic so it's really important to better understand what are the the implication of the climate stress test so whether they can help financial institution to reduce the exposure to climate risk and also whether it can help to to to finance the transition to the local economy so the contribution to the literature is is very clear I think it is the first empirical paper looking at at this issue overall the the econometric framework is very well executed so I will not have many comments on on this part and there is a set of very interesting results that contributes to an important debate which is how can we mitigate systemic climate risk is it by bank divesting from the polluting companies or is it by bank engage engaging even more to the polluting companies in order to support their transitions so I have several comments the first one is about the environmental risk measure so in the paper you are using the rep risk measure which is a measure of controversies so it is based on news and by nature it is more volatile than a carbon intensity metric which is more standard in the literature I think I think it can be a potential issue in the deep in the framework because in the deep in deep you take the average of the measure before the event so you are not controlling for potential change in the measure afterwards and also I think you want to focus on the long-term transition risk and due to the fact that the measure is quite volatile it might be more related to the short-term transition risk also it is not directly in line with the French climate stress test which is based on sector activities and then as you have mentioned they apply some long-term scenarios on the carbon prices so I would suggest maybe as a robustness test to use a more stable measure over time maybe even a completely stable measure like an extractive industry dummy or if you prefer you can use a dummy of the climate policy relevant sectors of batista or even the most on that and still more stable carbon emission intensity measure then the the results are very interested but not always intuitive so I would suggest using additional visualization tools in order to help the reader to to grasp your results and and yet to bring more confidence on the result overall so in the chart on the left these two charts are from mezzanine and angrienne so on the chart of the late you could represent the cumulative loans for treated banks and non-treated banks and maybe it would also help us better understand what the positive coefficient in the deep in the framework means because I think it can it can it can means either an increase in the lending to the most polluting companies but it could also means a decrease in the lending to the non-polluting companies or the less polluting companies and in the chart on the right yeah you could potentially try to run a dynamic specification so it would allow us to understand whether the effect of of the stress test is persistent over time then last thing to try to to to support your result even more maybe you could mention the IPCC figures that here you have the investment flows required in order to to limit global warming and as you can see the the flows should primarily go to some sectors which are the energy transport electricity and agriculture ones which tend to be bronze sectors so there might be some of course some heterogeneity within each sector but still as a as a as a macro fact I think it could help support your findings then the paper used a database of syndicated loans so I was wondering whether the effect could be specific to the syndicated loans and what would happen if you were using other type of loans or even securities instead so I was considering that there might be a potential trade-off I mean the treated banks could loan give more loans to the more polluting companies but at the same time invest less in the securities of the same company the rationale for this kind of trade-off could be a risk trade-off so the securities might be more risky in the short run due to the market to market but even in the long run in case of default there can be more refunding to the loans and and less refundings to the stocks for instance and there is also the influence trade-off that you are already mentioning in the paper so the loans and maybe even more the syndicated loans can create a long-term relationship with the company so create at the same time opportunities to engage with the company and it might not be the case for for all the type of loans or all the type of securities my last point is about the country effect because in the paper you are comparing the treated French banks with the non-treated European banks and there might be some specific effect for French banks because so in France we often have some headlines in the news like this like French banks tend to tend to invest more in fossil energies than other European banks so you are already doing a lot of things in order to control for this potential country effect I think you can do even more so for instance in the first equation why not running separate regression for the French banks and the other the other EU banks also the falsification tests you are using are very useful maybe you could run a battery of this test especially for the random stress test date and finally I was thinking that it might be interesting to try to run the diff in diff even if there might be a selection bias but between only the treated and non-treated French banks I just a small comment to to and I will be finished so regarding the syndicated loans I think you could add a bit more description about them I couldn't understand why reading the paper if the banks can revise the volume and the spread of the loans after having granted them so it would help the reader better understand what is in the database and what you do exactly in the regressions so overall congrats for the paper it is very interesting and as a significant contribution to the literature and yeah my main comments are running robustness tests with another environmental viable adding some visualization tools and perhaps enrich the analysis with evidence for other types of loans and securities thank you thanks very much Tristan for the concise but very rich discussion who you and I suggest we take some questions from the floor or in the chat before and then we'll be back to you at the end can I ask those who are willing to ask questions I see Klaus here and well there are four questions so I think we can stop here be very concise in framing your question as a question I try my best I know I can rush you on that Klaus that applies to others as well thanks on the last slide first of things are very good presentation in nice paper the last slide you say there is no evidence that banks actually terminate the relationship to banks to what was for brown and I wonder if this is really surprising because if you are a bank you don't want to push the customer over the brink cuts the funding and then you have down the road a default case and this brings me to a more broader issue maybe if distinguishing between green and brown is maybe not the full story because what we really want to do is we want to incentivize transition so it is very important that a brown company becomes green so you would need to look then at maybe research or transition plans and then you want actually to fund this transition and this could also be an explanation maybe for an earlier observation that you have an increased volume to the brown industry and maybe this is because these firms have been identified as brown they made transition plans and then they need additional funding to make this transition I don't know if the data might be an interesting way to to to look at if you haven't done yet thank you I thought there was there were three other questions from the floor they all they are all on that side then please go and introduce yourself please so Joseph Machinich from the Austrian Central Bank we did a similar exercise in 2021 and similar but not identical so very interesting what you showed we didn't have a look at at the data in Austria last couple of years but my gut feeling would be that banks haven't changed their lending or risk premier at all and I was wondering about the reason for that one might be that the difference between the Austrian and the French exercise was the bottom up and top down thing so I understand the French exercise was a bottom up exercise so banks were able to provide their data whereas we did it the other way around in Austria and to me it feels like it feels a bit like the difference between CT risk-weighted assets and leverage ratio so with risk-weighted assets you have the banks who are in charge of you know doing all those calculations and having their models and then when it comes to leverage ratios all of a sudden you see like what's really there and I was wondering whether we see a similar effect here so my suggestion would be that you do a similar exercise with a country that follow the top-down approach and see whether the assumption still holds true that banks are reacting well at all to a climate stress this and I would I would be happy to volunteer thanks I was actually to suggest you should join on that one I love the top-down thing by the way of course okay we have two more questions and Daniel Guchy ECB SSM many thanks for your interesting research outcome and linking a bit also to the previous paper on capital requirements I think in both of your papers cases you show that those measures capital requirements of stress testing are effective in achieving a certain prudential outcome in your case it's the increased interest rates for the borrowers that face transition risk but they're not so effective in achieving and let's say environmental policy outcome so in your case in inducing the borrowers to switch from brown to green technologies suppliers that's on a simple question should we expect prudential measures to even have any such environmental policy outcomes many thanks so make the mark sin from society general and my question follows on quite quite directly from the previous one because when when we do stress testing exercises we're basically looking at various future scenarios and I of course very much hope that the orderly transition is the most likely scenario but as I think the previous paper already hinted a little bit and what we see from from what's actually happening in policy circles it's not the most likely scenario and and I think it comes back to the question of of using stress tests or capital requirements to achieve environmental outcomes and it's I think what would be really really interesting to look at and maybe for the paper as well is to see what's happening in the the non-bank policy if we can call it that way because I suspect that if we have I'll call it non-finance policy if we can call it that way I suspect where you have non-finance policies driving a change in the real economy you would see that bank behaviors are also mirroring this and that could then be reinforced with capital requirements although I wouldn't advocate that necessarily but with the bank stress test but where the actual policy is not changing or the actual transition policies which drive the financial risk are just not moving why should we expect something to change in that case thank you under us I think you have again one point from the chat Paula I would like to understand better why the carbon price is lower in the initial years of the transition scenarios compared to the baseline scenario clear and short one William you have not too many minutes to address all of the five questions yes because we don't want to crowd out as we've stunted already my assume this yes I understand that I'm between you and really great keynote and also lunch so I will be really quick in terms of question about measurement of outbreaks and also other measurement for environmental risk from the discussion I think that we choose this reference also because it's less buyers than the one that firm sell report for example carbon emission which is from the CDP reports this is like firm estimated so we do agree that we would do robin it checks that also use carbon emission intensity which we have in the data more visualization I absolutely agree industry split I also agree because we see that some industry are carbon intensive but they are having some hope so we can have technology to transform them whereas some other like oils and gas would be much harder to do so we would expect if our results really true we should expect the effect that more to the carbon intensive industry that are more likely to transit the other type of loans versus six securities I think that probably we can only partially address this question we have firm level information on the total loans but for securities holding up banks unfortunately we do not have that great data to analyze but if you know other data source that we could actually take into account we would be really happy to really do this test for now I will move to the answer to the question from the audience so close questions on so we see no evidence in firms dominating the relationship with their suppliers so this is also maybe because of the distinguish between green versus brown suppliers in this sense we try to do a follow-up test where we collect information on loan purpose so let's say that we see changes in bank lending and maybe that these lending did not have any effect in the supply chain factor because the loan purpose was not really for the green transition so hopefully after we have this loan purpose in the loan level information we could be able to address that so the question by Joseph the top down versus bottom up approach of climate status I really like this point because I think that the ECB the first test of the ECB that has been done I think 2021 also a top down approach this means that the central bank collect information from the banking system and they use scenario and sell calculate whereas the bottom up approach like in France the banks need to also spend time and effort to estimate the losses so I do think that our results have something to also do with the fact that could be different scenario because these bottom up actually expose banks more to their measurement sizes mean that they need to put down a number whereas the top down I think that is helpful in terms of financial stability which is I agree that the central bank should focus on because we need to know the system metrics there but for the banks to change I think that there could be a common ground that we come with both top down approach and then bottom up approach so that is more on the bank side that they need to also expose themselves and calculate unexpected loss if carbon prices change and then the point by Daniel should we expect the potential outcome I think this is really a good point and I think that there's still a lot of debate on this because our view is that we don't think central bank should really have too much like targets because it's also not possible but what we agree with that if we can somehow align the incentives so central bank look at this from the financial stability aspect rather than because we need to see something like appear to be like supporting the green transition without considering the real level of risk but I think that if the central bank come from the financial stability approach and then later on because of the threats to the financial stability then the other policies that are getting bang would be more in line with with this goal and finally the point by Michela and I also agree that we have various scenarios and it's likely that we might face more disruptive and late transition rather than the orderly transition um so and then you you also have similar question than the than with Daniel's point so I will not repeat that part but about non-bank policies um so we have another working paper which talk a little bit about the non-banks versus bank interaction in measuring risk and managing this and what we find there is that because the non-banks they are less regulated so there's not much of like changes there despite the fact that a lot of assess manager declare to be green assess manager but the banks take advantage of this fact and they we find evidence that banks ship the risks this means that they sell off brown loans to these non-bank entities without having a carbon premium there so um and then the final question by Paolo from why carbon prices are so much lower in the late transition and also the disruptive more disruptive scenario compared to the orderly one um I'm sorry that I did not clearly explain it during my presentation so the orderly transition meaning that we take already action from now meaning that carbon prices can increase already from now on so we have a more gradual gradual increase in carbon price and it's also reflected in the fact that there must be technologies ready for the transition whereas the other scenario is like because we don't act anything now and we wait let's say 10 years later and because there if the effect of higher of like much more carbon emission lead to more natural disaster for example then we would face a much higher carbon prices and because of that the earlier years low carbon prices and the later years much higher one so thank you I hope that I address all the questions