 Mi ydych chi'n Paul Bates. Mi'n profesiol o hydrologi at Y Unifersiw Bristol, ac rwy'n meddwl i'r CGFI co-investigatau. Mae'n fflaesio i'r cysylltu'r seisio'r gweithio'r cyffredinol o'r ffordd ac ymddangos, ac yn fwy o'r gweithio'r gweithio'r gweithio. Mae'n gweithio'r panel i'r gweithio'r gweithio'r gweithio'r gweithio. Mae'n gweithio'r gweithio'r gweithio'r profesiol yw'r Gweithio Gennifor Cotto,oliios mewn Ysgologau Cyhoeddus, Mae gweithio i ddweud eich ddweud o'r ddweud yn cyfnodol yn gyffredinol, gofyn o gweithio yn cymaint o'r ddweud yw ddim yn ymddangos. Mae'r ddweud o'r ddweud yn cael ei ddweud, fel ar gyfer'r percipitation yn gweithio'n cerddyn, ond nid oes yn gweithio'n cyfredinol. rydyn ni'n gweithio i'r dysgu cyfnod ddeudon i'r cyfrifol am gweithio a'u deudon i'r cyfrifol a'u mobeithio. Roedd yn rhaid i'r cyfrifol o wybodaeth o'r ddysgu sy'n cyfrifol i'r cyfrifol a'u deudon i'r cyfrifol. First, it shows that the ground in the mid-latitudes affecting North America, North Western Europe and other locations, there's a maximum that is in the co-occurrence of extreme precipitation, and on the left you can see that in the green and yellow colours, and on the right in the red colours you can see where there's a high incidence of co-occurring extreme wind and precipitation. We need to understand the drivers of these co-occurring extremes. Are the extremes caused by common drivers so that they are more likely in certain situations, and this can help to estimate the risks from them. We need to understand this better to quantify both the present-day risk and also future risk. Changes in the driving weather systems in the future might affect the risk of getting co-occurring extreme events. I do this by thinking about different weather systems. This is just an example. It hasn't come out correctly. I think about extra-tropical cyclones, frontal systems, and warm and cold fronts that we get very frequently over the UK and Europe, and also thunderstorm environments, and how these different weather systems combine to change our risk of compound events. This is the wrong figure on the left because we've gone from Mac to PC, and so the figures don't come out correctly. What I wanted to show you is that when we think about extra-tropical cyclones and fronts, what the figure on the left is meant to show is that cyclones and fronts combined together and also fronts on their own are the most common weather type for producing co-occurring precipitation and wind events in winter in the middle attitudes. So in the area of the UK and northwestern Europe, co-occurring extreme precipitation and wind is mostly associated with extra-tropical cyclones, and another study on the right, the purple outlined box shows the proportion of winter events associated with extra-tropical cyclones, again, and this shows that a very high proportion of co-occurring precipitation and wind events are associated with extra-tropical cyclones. So the various different ways of looking at this from a weather system perspective, but they kind of give the same answers. So is it that the wind is causing the precipitation or vice versa, or is it that they have a common driver? So the research that we've done shows that actually the proportion of extreme precipitation events that also have extreme wind events is much larger for these extra-tropical cyclone systems and lower for more tropical systems, which is different to when we have, say, a wave wind compound extreme where the weather system type does not influence the rate of co-occurring. So that suggests that the middle attitude weather system is actually the driver of the, is the common driver of the extreme precipitation and wind compound hazard. So research that's ongoing, I put up this figure on the right just to illustrate the type of weather system that we're talking about. So this is an extra-tropical cyclone with the low pressure in the centre warm fronts and cold fronts. And so we're doing work to try and understand more about the characteristics of the driving extra-tropical cyclones so that we can better understand the potential risk of these events from the weather systems. So we're doing that looking at a storm-centred point of view, which is what's shown on the left, to see where these co-occurring extremes happen within extra-tropical cyclones. And then hopefully we can find out more information about other characteristics. So is it the strongest storms? Is it when they're coming from particular directions? What is the impact on the, what causes the higher risk of compound extremes? And we're also doing research looking at the impact of climate change on these co-occurring hazards in the future and some research that's not published yet but thanks to Colin Manning. It shows an increase in the footprints of co-occurring extreme wind and extreme precipitation in the future. So that's what the panel on the right shows, an increase in the frequency of these co-occurring events. So we want to understand better the characteristics of the storms that drive these co-occurring hazards so that we can estimate our risk both in present day and in the future. So that's me. Thank you. Thank you very much, Jennifer. So now we move on to Dr Hannah Bloomfield, who's now of the University of Newcastle as of the start of this month. Hannah was up until then the CGFI postdoc at the University of Bristol working with Len and I and Hannah is going to move now from thinking about wind and precip to thinking further into the impact space and about wind and flood and actually wind and loss. Hannah, over to you. Great, thanks. Thanks everyone for coming to this session. Yeah, so as Paul said until about 12 days ago I was the CGFI wind flood postdoc and we've talked to you about some of the stuff that we've been doing over the last 18 months. So I know it's quite a mixed group of people so in case you were kind of, you know, it's late in the day kind of wondering what is one of these wind flood events in reality and maybe what does it mean for you. So a good example of one of these might be storms Dudley Unis and Franklin. So in February 2022 we had a series of storms coming through. So we've got some pictures of things that are a bit broken associated with these storms which potentially caused you all quite a lot of headaches. But also you can see weather maps here and you can see overlapping regions of different weather warnings. So particularly for storm Unis and storm Franklin here you can kind of see overlapping regions of a wind rain warning and maybe these kind of winds regions. So for us a compound weather event in this context is when you've got these two kind of hazards overhappening in an overlapping region or maybe at the same time or maybe within a given time period. So when I started this postdoc first job was to read lots of papers and there's actually a reasonable amount of literature from the academic community thinking about wind flood. So it's like we plan this. Some of these papers are the ones Jen was talking about earlier by a PhD student Laura Owen. There's a lot of work thinking about very much weather variables. So thinking about like wind and precipitation, you know, this kind of climate science perspective. But I was then employed in a hydrology group. So really if you're thinking about flooding and you want to move towards impact, you should maybe start to think about how that precipitation impacts river flows. So there's a few studies kind of in this camp. John Hilli is here today I think. But when you really go down to these maybe damage relevant metrics, stuff that you might be able to implement in your kind of studies, I think Sea Bears was a prime example of this. Like what papers do we write that maybe have usable metrics for you if you were conducting that kind of exercise? There's not many, right? There's two here. Well done John. And they all use various different statistics. It was quite difficult to try and synthesise this. So I spent some time in CGFI trying to synthesise this work across all different time scales and lots of different impact measures. So you might have seen this already, but this is our CGFI Wind Flood Correlation Explorer. So if you're bored of me talking, you can scan the QR code and play with it. But what this allows you to do is to see all the results from the papers we've written. So you can choose a timescale of interest. You can think about either here we've got weather variables, we've got gust versus precipitation. You can also think about some more impact focus variables. So as part of the project, we worked a lot with storm severity indices, which are a common way of the insurance community thinking about wind damage. We also developed a novel flood severity index. So a similar thing, thinking about river flows exceeding critical percentiles and how that may then relate to potential for damage over an area. So if you play for this for a long time, you've kind of accidentally done some science and you can then come to these conclusions. So you can see from it some example maps here. This one's looking at gusts and precipitation. The values are different across Europe and this kind of relates back to what Jen was talking about. Depending on where you are in Europe, the drivers of what causes your wind flood correlation and the strength of this are different. So the numbers are highest, like closest to a value of one down the west coast. This is where we get most storms interacting with the coming off the North Atlantic. You can see how these two maps are different though, right? So once you go gust versus river flow, the results are different. And generally, as your metrics become more impact focus, the correlation values do drop, but we still see significant correlation between wind and flood. Why is this important for you? Well, generally when you're modelling these hazards, these are done as separate models, right? You have a cat model for wind and a cat model for flood with providing some evidence of why you might want to stick them together. In terms of what drives this, Jen has talked a bit, but in the paper we unpack for different timescales from this maybe storm centred start. It's all storms really. It's just about how many storms we're getting along the timescales. The hydrology starts to become a bit more important. So if you're going to have a big compound event, it might need to be conditioned by having particular soggy catchments, so they're not able to absorb any more water, and then you're more likely to flood. On the very long time scales that might be related to the position of the jet stream, phases of critical modes of variability, like the North Atlantic Oscillation, and we're about to submit a piece of work that unpacks the meteorological drivers of our very extreme storm severity and flood severity index work to show that basically there are particular locations of cyclones that lead to the most intense UK compound events. The very final bit, I guess, is although I'm doing more impact focused work here with CGFI, it's still a little bit removed from what you guys do every day. So the final thing I was doing as I left, and still now because I didn't finish it on time, is working on the potential to do this toy loss model. So we've been building event sets, a more traditional input to a catastrophe model than these time series I was showing before, but we've built the first kind of compound wind flood event set where all the sites in all these dots are different catchments, river catchment sites around the UK, and we can build a statistical model that conditions so you've got extreme events based on both the wind and the flood being extreme at different sites. We can combine this with open access, exposure, and vulnerability data, which I caveat is very difficult to find, to get our first kind of estimates of how compound losses may compare to these losses from individual models. So this is work in progress, maybe come to the forum next year and we'll tell you the answer, but yeah, hopefully leading towards more impact focus work. The last thing I'll say before I hand over is if this kind of thing is interesting, please check out a CGFI event we have a recording from for. This was a Grand Challenges meeting where we spent quite a lot of time talking about compound wind flood as well as some other big issues, and this was in combination with the Royal Met Stock Special Interest group. So that's all from me. I'll pass on to Jess. Thank you very much Hannah. So now we're going to move from the science of wind and flood risk to how those are viewed within the industry, and I'm very grateful to Dr Jessica Turner from Guy Carpenter to give her the next talk. So Jessica is head of the Catastrophe Advisory, a Guy Carpenter who is one of our most experienced cap modelers and has previously worked at Willisbury, RMS and Lloyds Banking Group as well. Jessica, thank you very much. Thank you today. I hope you've been having a great session so far. So as Paul said, I mean I am a meteorologist by training actually, but I've been spending the last 15 years in the insurance industry and cap modelling industry thinking about how these hazards impact insurance, and so that's mostly what I'm going to focus on today. But I wanted to first start and give a little bit of a framing. Now I'm going to stick here to the mid-latitudes like the previous two speakers, I think tropical cyclones is kind of a slightly different issue for thinking about Europe and about mid-latitudes. When we talk about wind flood correlation there's actually some nuance there, so there's different types of wind flood correlation, and those have impacts, they have financial impacts. So the first one we should think about is really are we talking about seasonal correlation, or are we talking about event correlation? So seasonal here, let's say for over the winter season, you know, from November to March, it's windier and wetter than usual, but not necessarily at the same time, meaning there's a series of low pressure systems coming through, sometimes you get wind damage, sometimes flood damage, but they're not co-occurring, and that's important because that's going to impact the profitability of an insurance company. When we're talking about event co-occurrence or correlation, it's when the same storm, say Eunice in one of the previous examples, or Desmond which caused the Boxing Day flooding in the UK, that's when you have wind and flood simultaneously, and that's more important for reinsurance, so you're more likely to be able to claim on your reinsurance program if you have an event. So for us in the industry it matters if we're talking about seasonal correlation or event correlation. Another thing to think about is are we talking about the winter season or are we talking about the summer season? Because you can have wind flood correlation in both cases, so in the winter this tends to be caused more by large-scale synoptic frontal events and you've seen a few images previously, but we can also have wind losses in Europe associated with convective storms in the summer, so these will be small-scale convective storms causing hail, straight-line winds, tornadoes and flood losses from extreme precipitation. So with those in mind, I'm now going to talk through a little bit about the losses, so oh I did have one point to make, sorry. So as Hannah said previously with the exception of tropical cyclones and coastal flooding where the correlation is generally taken into account, most of our contestry models assume independence of wind and flood models, and I've said most kind of as a hedge because I actually don't know any models that do that, one might exist, but in my experience all the vendor models have those as completely separate, the wind and the inland flood. And there are some good reasons for that and some less good reasons. I think a good reason is that the vendor models would say well usually the storms that cause really damaging winds over Europe are fast moving and they tend not to dump a lot of rain in one location whereas a big flood event will come from a stationary front sitting over a location for several days. But I think we've seen it's not, they are not 100% independent, we know that from the hazard already. So I'd argue that it's something we should start to think about and take into consideration. What I want to do is talk about hazard correlation as seen through the losses, I think we've seen that we can see that in the hazard data, but do we see that in real world losses? So what I'm showing you on the screen here is UK household losses as provided by the Association of British Insurers. I removed the scale because this is data by subscription, but it sort of doesn't matter because what I'm trying to do is show the correlation between storm and flood, where storm is in the purple and flood is in the blue. Now you've got winter season losses on the left hand side and summer losses on the right hand side, so this is a seasonal metric. You should also take into account the fact that the data is untrended, so I haven't taken into account changes in currency or the built environment, but because we would trend storm and flood in the same way, I think it's fine to leave it out because we're just looking at the correlations here. And if you do a really simple correlation just in Excel, that's all I did. You'll see that there's decent correlation, you've got 0.5 for the winter season and maybe a surprising 0.8 for the summertime losses. So that's interesting, that's an indication we are seeing correlation between these perils in our losses. But you do have to be careful because the ABI defines storm as gusts over 55 miles per hour or rain above 25 millimeters per hour. So we don't really know a loss adjuster on the ground might see precipitation surface water flooding and might be calling this storm. So there is probably some contamination in the data. What about real world examples? I wanted to talk a bit about the summer of 2021, which was really an extraordinary summer for continental Europe, which was characterized by extreme hail, wind, tornado throughout essentially all of June and even into July. This was a really interesting meteorological situation where there was a heat wave over continental Europe, very hot. There was a stationary high over the Balkans, which set up a jet stream coming from the southwest and one low pressure system after another brought cold moist air with a cold front over the warm continental Europe, just driving convection cell after convection cell with heavy precipitation, straight line winds and hail. I've written down all of the low pressure systems because I would never remember them all. We had Stefan on the 18th of June, Fanonant on the 19th of June, Ulfert on the 20th of June, Volker and Wolfgang, the 21st to 25th of June, followed by Zero, the 28th to 1st of July. And at the end of that extraordinary series, there was 3.5 billion euros of insured loss in continental Europe. And then the whole series was capped off by extreme flooding in July by the storm burned. And this was the famous German flooding, which happened in 2021 in Germany and in Belgium, where unfortunately more than 180 people lost their lives. So here's a real world example where we see we've got hail, we've got wind, tornado flood, all happening at the same time. The flood, according to the cap models, would be modelled completely separate from the hail, the wind and the other perils. And clearly that's not correct. What about modelling studies? So a few studies have been undertaken to use cap models to investigate this issue a little bit more. So I'm going to highlight one that's undertaken by the Bank of England, with some academics and also the modelling firm Verisk. And here they're looking at seasonal correlation. So what they're doing is taking the output, the simulation output from a wind model and a flood model, and they're matching up the years such that they're forcing correlation, so essentially resampling. And they assume either a 20% correlation, which is the bottom two bars, or a 40% correlation, which is the top two bars. And you can see the impacts on the net loss, an orange and the gross loss in the blue, on the 200-year return period annual aggregate loss. Now these numbers are relatively moderate, you know, we're talking about 3% to 6%, and maybe that doesn't sound like a lot, but the 200-year return period annual aggregate loss is the solvency capital, which must be held by firms. So a 6% increase, you could be talking about easily hundreds of millions of pounds for a firm. Event correlation, so this is a study that Guy Carpenter did specifically looking now not at seasonal correlation, but at event correlation. And the first thing we did was look at era five reanalysis, which is the historical observations from 1979 to 2020. And we calculated the conditional probability of extreme precipitation, precipitation over the 98th percentile given daily maximum gust exceeds 20 meters per second, which is generally considered the threshold to be damaging. And we found some decent correlations between the two. Some areas, particularly around topography, the correlation is very high, but over much of Europe you're looking at around 0.5. So we did the same thing as the Bank of England, we just did a resampling of the wind model output and the flood model output, but here we did it on an event basis rather than seasonal. And the picture we found was a bit different. We found pretty small increases actually in the event losses when we correlated the wind and flood. So you can see from the table ranging from 0% to plus 3% depending on the return period in the country you look at. And the reason for that was that the wind losses so dominate Europe, the magnitude is so much larger than flood losses that even when you correlated them on an event basis, the impact was relatively modest. So my last slide is just to talk a little bit about what are the financial implications of the wind flood correlation. The first two points, increased property damage and increased business interruption. The last two modeling studies I showed you assumed still independence of vulnerability, meaning even when the event co-occurred, you didn't get the compound effects to the vulnerability, to the damage ability of the property. And of course that's not true because you might have heavy precipitation at a location which wouldn't have flooded the building but for the roof being blown off. So that's an example. And of course increased business interruption if you've got a flooded property but you've got trees down blocking the road, you can't get to it and you can't make repairs. And then my last three points are around sort of the profit and loss daily management of a business. So if you have correlation of wind and flood, you've got increased volatility in your financial returns which investors do not like. You might have increased capital requirements as I've shown previously and you can have poorly structured or priced reinsurance. Okay, thank you. Thank you very much Jessica. And last but by no means least, something we've already heard from a little bit today and probably needs no introduction but I'm going to do it anyway. Rowan Douglas CBE from No Longer, Willis Tannas Mawson. That's right. I didn't know I was making a presentation but I will say a few words. Say a few words on your thoughts on this topic. No, thanks so much Paul. You said me to say this once before but my biggest claim to fame was doing my Enfield Viva the same day that Paul did his PhD Viva in Bristol just a few years ago. Yeah, absolutely fabulous presentations. I don't have a presentation and we'll make some comments on the panel but I guess my background is re-insurance way back when I suppose perhaps to the general audience there's perhaps just three comments I'd make and it's been fabulous to see this focus on such a driving peril or hazard of wind and flood and accommodation and quite rightly there is a tremendous focus on understanding the hazard or the peril and you know it's a it's a climate it's a climate conference but when understanding that the risk to individual properties or crops or whatever when we when we go through risk modelling it's also critically important to understand the attributes of those exposed assets or populations or even natural assets probably the biggest revolution the insurance sector had to go through and why you have to spend far too long filling out your insurance application form is to capture the attributes of those exposed assets and then critically and they were touched on in a number of the of the talks by Jennifer Jessica and oh my goodness me I've had a mental blank Hannah, excuse me, is to understand the vulnerability functions that actually affect those assets to those particular hazards and if you want to understand the the secret source that often changes so much the outputs of expected losses and ultimately to the point that Jessica made and maybe not everybody picked it up insurance companies do have unlike all other financial sector institutions insurers and insurers re-insurers do have to hold capital from the model so we talk a lot about you know will there be a capital charge against some of these risks for insurers there are and the outputs of the models drive exactly how much capital an insurance or insurance company has to hold and so often it's the it's the effect of the vulnerability functions and assumptions that change the biggest thing with with the outcome so but for non-insurance institutions capturing that data on their exposed in exposure in their portfolios is often the most challenging thing so second thing I'd say and it was it was it was picked up as well is we will have to think very imaginatively about the triggers and thresholds which we and imaginatively the triggers and thresholds we develop that will allow whether it be financial institutions or even countries I was in Bond last week at the at the UNFCC summit talking about loss and damage and what are the sorts of triggers we could develop to enable us to impute what the expected losses are from in this case wind and flood losses to economies to make sure that they they understand what those key thresholds are and manage to them and try and become more resilient but ultimately can receive appropriate entitlements to cash including from the global north when events occur and we have to spend a lot of time working through what those thresholds could be to provide those triggers. A third point I wanted to want to say was it is vital that as I mentioned at the beginning of today's conference whatever financial section situations we come from or public sector body we become more like geographers as well as economists and we get the physics into finance and it's important that and Jessica your journey from sort of science through to reinsurance little time at banking and and and back into the reinsurance world was instructive. Every bank every investor will have to hire a hydrologist as much as meteorologists as much as economists but I do sense that it will probably be the requirement to ensure these sorts of risks far beyond our existing requirement unusually in the UK to require flood insurance and other natural hazard insurance for mortgages and other financial commitments. I sense it'll be the insurance and reinsurance sector quite broadly defined it becomes effectively the risk management valve through which other financial institutions decide whether they will take on a alone or or some other financial commitment because I think the requirement and cost of modelling and human personnel required to truly get to grips with flood and wind risk will mean that I think that that sector which is already we did some back of the envelope statistics a few years ago and worked out the insurance and reinsurance insurance and reinsurance sector around the world already spends and this was three or four years ago two billion dollars a year on risk modelling catastrophe risk modelling that includes all the hundreds of millions of dollars they have to spend on catastrophe risk models maybe maybe not quite so much if if tom has his way but also all the personnel and all the other requirements it was about two billion dollars it's a massive investment and that wasn't necessarily with a forward gear looking into the future of climate that was just looking at current levels of extreme risk so I think it's going to be unlikely that all other sectors want to make that same or even larger commitment and I think the CGFI is perfectly placed to allow the banking and investment sector to take full advantage of the commitments that the insurance insurance sector will have to make but also add facets and components to make sure that the the gearbox that is used can translate this to the decision making in a bank or an investor as much as it has in the reinsurance world and even in that world we're having to understand how to put a proper forward gear onto the gearbox the gearbox is basically right but it's um it's it's currently got lots of reverse gears and it needs a few more forward gears so um there we go that was my painting picture with words so now I'd like to invite the panelist to join us on the stage so we're on uh not so fast uh we need to back up here um and we're going to I'm going to hand over to Len to do a little bit of Q&A. Okay hi everybody I'm Lin Shafry from the University of Reading the National Centre for Amnestyx Science um so how's our panelists their way to the stage? I'll just say that we've got roughly about half an hour for questions and answers in this panel session um I've got a couple of prepared questions to ask the panelists and then we'll take some questions from the floor so you've got a few minutes to think really difficult tricky questions that you've always wanted to ask about so I guess to start off with that I guess um I know one of the things that came from your presentation when you sort of showed the the kind of the the the published literature just how recent all those studies are I mean it kind of brings home just how recent a subject this is it's kind of over the past decade we've started to understand a lot more about correlated risk in Europe in terms of wind and flood um so I guess that I kind of wanted to ask everybody on the panel um from your point of view where are the big knowledge gaps there's obviously a lot more that we could learn about about wind and flood risk there's lots going on in the science but clearly a lot's going on in financial institutions particularly in the insurance industry so from your perspective where are the big knowledge gaps and how are we going to make progress so maybe maybe Jen I guess can we start with the meteorology sure um yeah so I think I think there's still a lot so so compound hazards and compound events of kind of really had a lot of attention in the last few years um in various different places and and not just the compound and wind flood events um but I think one of the one of the problems that we have when looking at compound events is the data so because when you're looking at extreme events you are already limiting your data sample and if you want to look at the top 1% strong as wind events if you then want to look at those joint width um flooding events you're really limiting the data that you can use and so I think one of the one of the problems is really understanding how um how these events do correlate um with such a small data sample and so I think that's um that's a gap um so making use of large ensembles of um model data and we've actually had a project with Guy Carpenter to do this using seasonal forecast models um to generate more like a larger event set using physical models um yeah so that from the meteorological point of view I think that's that's one of the issues is looking at the weather systems and the lack of data um yeah and making use of physical models to actually make some progress on that yeah I guess to extend that if you if you make the assumption that you can get to the weather data if we say if we take that one as a given and okay okay we've managed to find a nice large ensemble that we can do some analysis with I think one of the things I've found most difficult and with the corona showed is trying to get towards the impact so with these correlated hazards like just generally wind and flood are very correlated right generally when it's a bit windier it might be raining as well like just winter weather versus summer weather but in terms of the events that actually cause damage um we're into the tail and we don't have that many industry observations at this point like um I showed some very dodgy looking vulnerability curves that I've been making but to find open data we can use in academia to try and calibrate something you know and understand like for a given event um you know we've got this event we know what we think happened meteorologically was there actually a loss um that's that's incredibly difficult for us I think a big gap um in the modelling yeah I think um I'm very grateful for the studies that have come out recently I think it's been very interesting and I'm glad it's a topic that's being explored I think it's less of a problem about limited data sets because in a way that's what catastrophe models are for I mean they're built to look at extreme events I sort of think we've got enough on the hazard side that we could build correlated catastrophe models and I think do a decent job I think the problem is really implementation the desire to do it and I think it's it's because historically developers have seen wind and flood as not being that correlated when it comes to losses um so maybe some of these studies are going to help motivate change there and I also think it's just it's already hard enough to build a flood model or to build a wind model you know it's 15 PhDs working for five years on these things and I do think they're they're conscious that that needs to be done and that's maybe a next step that's coming I mean another thing is that traditionally meteorologists and hydrologists are coming from slightly different academic backgrounds and they can tend to be a bit siloed I think uh when it comes to working together present company accepted I have absolutely nothing useful to add so I'm going to move to the next next question apart from to say of course let me focus on the correlation between these two types of peril of course uh if you if you're looking at an international or even global portfolio another dimension of correlation which will be touched on I'm sure in another panel another day is is that we've looked at it for many years haven't we the challenges of understanding the correlation spatially in any given year the sort of teleconnections it's a different subject just to add further but we're talking now here about the correlation very importantly about um uh if you like co-located perils the other challenge that we'll also have is to understand what are the potential correlations are around the world both for this sort of peril but more broadly but uh no I'll leave it there so I guess moving on to something that kind of CGFI is thinking about in great detail is on climate change uh I mean from your perspectives I mean do you feel as the climate change and it warms will compound risk become a bigger ratio would it be something that perhaps we'll consider less is it something that's going to come higher on the research agenda one of the studies that I showed that we did at Guy Carpenter we also we didn't look just at the historical observations in the correlation we also used CMIP6 uh and looked at the same correlations and we did see that you know storms were tending to get wetter as I guess you would expect I mean that's kind of an obvious consequence so it did seem to to drive up the losses a little bit it wasn't as much as one would expect but I think what worries me more from the climate change perspective is more about the attritional losses it's more the seasonal correlation rather than the events correlation and we're just seeing more smaller events but they add up yeah yeah just to echo that it's part of the study we talked about there um where we're thinking about these correlations between these storm severity and flood severity so we did that using the uk climate projections the uk cp18 data set which is a single model rather than a sweep of future climate models but that one showed if you look at your most extreme compounds um a fourfold increase in the future which is mainly driven by um the flood severity index getting much higher so it's just generally much wetter um so we had a similar number of these wind events but just more floods so more likely that they would then be a compounding similarly with the seasonal timescales that then adds up to just yeah catchments generally being wetter yes yeah yeah unpacking the anti-seeding conditions I think is really interesting for this um and seeing if we could maybe understand a little bit better when something really bad might be coming yeah I think the evidence is really building that um co-occurring the extreme events are going to become more common within particular storms um so there's also the shift of the storm track that's going to impact where these events are occurring so some of the some of the impacts will depend on where you have the vulnerability or the exposure I mean um so um but there's there's quite a lot of evidence to show that um the storm tracks are going to intensify over northwestern Europe and so um and with that increased precipitation intensity and larger footprints of strong winds so I think I think these compound events will become more important in the future no it's incredibly exciting to see how this higher resolution climate models are just giving us this sort of crystal ball into into the near term that's a medium term future I just highlight that the intellectual challenge is is to marry climate models they may be quite high resolution but compared to risk models they're very low resolution and and the magic is to use higher resolution climate models to inform our our risk models and our our stascastic event sets and the rest and and and that's a sort of work I think is so critical that the cgfi can can can can bring together so but of course what will also drive the loss it's not quite such a fashionable topic is how our expose how our exposure uh evolves maybe not quite so much in in more mature economies and geographies like the uk but certainly in in developing an emerging economies it's the changing rapid nature of of urbanisation as well as changes in agricultural use that are actually having a massive impact on the level of loss of course the changing hazard is also critical to and somehow we've got to marry all these elements together interestingly when we've looked in the united states for flood alone the the losses due to population change the increase in loss due to population change it's about four times that of the increases due to climate interesting so even in developed economies which are growing rapidly but that is so pants up yeah I think that's a really nice segue into the question that I was going to ask now before I take questions from the floor is we've kind of talked here quite you know specifically kind of european and uk wind and inland flooding what are the compound hazards or compound risks should we be looking at what what what should be on our radar I think um thinking about in the future future risks with climate change compound coastal flooding is one of the particular issue and with intensifying rainfall and more flooding inland plus storm surge higher sea levels that's compound coastal flooding is going to be an issue um and there's been research on that in the UK climate resilience programme and looking at a different coastal compound coastal flooding events around the UK and but then also heat and humidity and heat and drought which impact on agriculture in particular and and again in the UKCR research and there's there's been um projects looking at that and the impact on potatoes and cattle and things like that so yeah I think there's there's other definitely other compound events that are going to be important yeah I think for winter compounding so we're just starting a project with John Hillier who's maybe sitting somewhere Isaac there hello John looking at lots of winter compounds so potentially ones that stick out are maybe cold wear and cold dry events for causing disruption maybe not large loss and damage but potential to disrupt rail sectors so I'm an energy meteorologist by training so I mean we're talking about net zero a lot today in theory if you want to keep the lights on in a very sustainable way the biggest compound event for the UK is cold weather and low wind right so it's not damaging in the traditional sense but if we do move towards something where you have got actual carbon credits that work and that's a very low wind event for a long period can seriously impact the finance sector. I think in the future with an increasingly volatile hydrological cycle you could get many more years in which there are both co-occurring drought and flood episodes and that could be interesting for some insurers because with dry weather we can become subsidence and you know agricultural losses are also problematic so seasons or years where we get co-occurring drought and flood could well be something to look at. Yeah that swing between drought and flood particularly soil compaction and kind of runoff and things like this is a whole stack of issues though. I'm going to see whether there are any questions from the floor oh gosh lots of hands going up. Quinton I think you were first there and I think you might be in John and I'm sorry I missed oh gosh there's lots of hands up. Hello Quinton Rae who won investment management now I've actually got two questions so you can have a choice if you don't want to one for Jennifer and one for Hannah. Leaning on some rather ancient geophysical fluid dynamics knowledge so take your pick. Who'd like to start? Jennifer or Hannah Grawl. What's the question? Well I'll start with Jennifer's question. It made me think about energetics so you're asking about the cause for the co-occurrence or both wind and precipitation. So I thought to myself maybe the common cause is to do with the state of the energetics of the atmosphere because obviously strong wind carries heat and latent heat of evaporational condensation carries heat. So maybe it's I mean this kind of goes back to how I did geophysical fluid dynamics which is if you imagine some kind of a surface at a constant latitude going vertically up and you ask yourself could there be you know what would maximise the heat transfer going through that surface and how would the atmosphere set itself up like that as your cause for the common cause for the two things. Does that sort of make sense? Is that an idea being explored? So I guess when it comes to our extra tropical cyclones and low pressure systems we've got a fairly good idea of where we expect to see the strong winds and also the precipitation that is the structure of the cyclones and the frontal systems you get rainfall where you've got uplift and so sometimes the stronger the winds the more the uplift and so there's that connection but also to get the heavy precipitation you need a lot of moisture so where the storm is coming from is really important but yeah understanding a bit more about the the characteristics specifically so why do some storms produce extremes and others don't and is yeah still a topic of current resources. Yes you've obviously got to get your moisture from some moment. I'm thinking mostly sort of energetics moving north from the south. Okay well the the question I had for Hannah was about co-occurrence and floods droughts because you're looking at the correlations and what struck me was you get a situation maybe where you've had a drought because it dries out flood the soil lot and I think that Hannah almost touched on it and I don't know what it's been extent it's been researched but then obviously if you get heavy precipitation that follows which you could easily imagine you've got a sort of flash flood situation so you talked about the the memory and the catchment and that seemed to I mean obviously that's perhaps a slight tangent from your work but I wondered whether there was anything on that that was sort of coming through. Yeah absolutely I'm definitely going to defer this to my hydrologist ex boss as I'm a climate scientist but seems fair. It's a good point though I think like a lot because it's winter a lot of the time we were assuming with what I was talking about there that you are getting storms particularly in these clustering periods like Desmond was part of the cluster right and then you get this tipping event where finally everything is saturated but yeah Paul do you want to say anything about winter flash flooding? Yeah so combined switches between droughts and flood is a current active topic of academic research. I'm not aware of the studies that are ongoing they're just starting to produce information and putting it out in conferences so there's nothing really I can direct you to at the moment but there's a growing awareness of it and an issue and there's research ongoing at the moment but yeah so you're absolutely right and as Len said you know drought conditions very hard compacted soils you could get very extreme rainfall afterwards you could get a lot of runoff interesting as well and guys probably won't appreciate this but where you get wildfires actually wildfires create hydrophobicity in soils so water runs off more from burned land than from natural terrain so if there was a possible outcome it's in ground wildfire yeah so if you have if you have wildfires over large areas as you might in California then one of the risks that you might be concerned about in the winter season afterwards is increased runoff because of hydrophobic soils okay thank you very much everyone so I think I think John Hilliw was next and then okay thank you quick question how would you define event what time frame and why it's going to be fascinating for me to hear what's most relevant interesting to you from your your viewpoint of perspective and that's to any one of the time frame case it should answer it or say I laughed because I'm putting together an event from the ISCM the international society of catastrophe managers this autumn to deal with exactly that so from a meteorological perspective it might be quite different from an insurance perspective you know we have contracts and there's wordings in those contracts and it is a matter of great debate really so in some cases it's atmospheric disturbance so you argue about that afterwards we had some clients that I mentioned that an entire string of severe weather over the summer of 2021 we had some clients that wanted to argue it was one event because it was one atmospheric disturbance if you like which was the southwest early jet and so they thought they should recover all of that from their reinsurance treaties some people want to go to named storm even in Europe which is pretty problematic lots of contracts just define it by number of hours so we have what's called an hours clause 168 hours 504 hours or something and that takes away a little bit of the subjectivity but I'm sure from a meteorologist it's a very different perspective well yeah I mean the events that I'm thinking about are are individual storms and or individual like grids based and co-occurrants but I think when it comes to the research I'm happy to be guided by what's useful and for for you and others yeah I have no fixed view on what that would be and if I could add the catastrophe models have different event definitions within them so for convective storm for example there's one vendor who every single day is a different event and then a different vendor aggregates events together into kind of micro macro events rather based on what they think are the atmospheric conditions so even practitioners who are using these models you need to be very careful to you to make sure you understand what you're modelling and what they think an event is and is it the same as your contract says is an event another use on events yeah I think for me I couldn't decide when we were doing the paper so I did everything from one day to 180 days so you can pick your favorite but when we were talking earlier about having small data samples and when you're in the extremes I think my from a climate science perspective my preferred event definition is a day maybe you'd accrue a few days rain before that or like the maximum 72 hour gust over a period but that just gives you the most number of points right if you start doing average over five days and you only got extreme events you run out of data quite quickly so although from an impact perspective that might be much more useful yeah we need guidance to do that thanks very much I think the next question was in the middle here hi um Albertine from Columbia through your investment sorry my question less academic probably more big scale it was to you Rowan you mentioned um you mentioned that the insurance sector might become I think you said the valve to which most finance so you know I work as a manager but as a manager where banks can maybe understand and quantify this better um my question really is just could you explain how you see that working how you can get involved with that and how some of the information that is available with the insurance sector can be applied to the longer time frames and the broader asset classes that we have to look at um asset managers sure absolutely and I'm not for a moment suggesting that banks and asset managers can't understand this risk I'm just it's just a case that necessarily insurers and reinsurers have had to over three decades now since the early 90s have had to invest so much money and it's almost changed the demography of the industry hasn't it so it would be a shame not for the other sectors to at least piggyback off some of that as well as add to it so I'll try and come up with an example and then but I don't absolutely know the answer to your question but I want to sort of illustrate a pathway so those of us who haven't come here from abroad we sort of take it for granted that we need insurance when we have a mortgage it's uh there's actually about seven acts of parliament or other legislative mechanisms which drive about 90% of UK insurance so the fact that we all buy uh or have to have insurance when we have a mortgage is because there was a clause in an act of parliament in the 1930s which allowed banks to require people to have insurance uh that's just one example it's just one little clause I actually was so boring I looked it up because I was thinking well it's going to be the same with climate there's going to be a few clauses and a few obscure contracts to actually drive the market so that's why if you went to the US or many other countries in the world you simply don't need to have insurance to have a mortgage so actually uh the bank may well have a liability doesn't it understand in the US if your house gets blown down or or flooded or whatever you can literally hand the keys back to the to the landlord sorry to the to the bank and it's their problem in the UK it's your problem so so we should never it's it's a bit like a sorry but there was a presentation downstairs when people talked about people withdrawing wildfire coverage in in California right now and and partly it's one of the reasons is because in the US the rates of people can charge for insurance are regulated by elected insurance commissioners who put restrictions on what people can charge so uh it's not always just about a fear of of of the risk it's so we take for granted the environment we're in here in the UK and assume that's how it is everywhere else but going back to the exam question so effectively I'm going to be I'm going to probably say things that shouldn't be recorded but in a way a bank doesn't have to worry in the UK quite so much as in other countries about his exposure to flood and other risks in in uh in the UK because fundamentally the liability is on the person taking out the mortgage and the insurer should pick up the town it is the in that case the bank is quite understandably subcontracting the decision of is this house too risky to insurer or too risky to put to to give a loan to because it should be the insurance company's job to decide whether that is insurable and at what price and in the UK because insurance is always a public private partnership we developed something called flood read to help you know 500 000 homes in him in exposed locations but put some conditions that supposedly you can't build new homes in flood so in effect and someone and actually insurers withdrawing coverage from certain places is sometimes perhaps the ultimate risk signal they may want to make money in places but if if somewhere is too risky to insurer adequately it's actually saying something else that this place is too risky perhaps to have a home so what I'm suggesting is that in that example UK mortgages banks are effectively using insurers as that valve deciding where they can provide loans in flood risk areas I don't know how that could work in other domains of of finance but I think it would be quite exciting to see where we could use that a to make sure that we have much bigger risk sharing mechanisms to share these risks as they propagate through the financial system but also desperately to integrate the platforms and technology and you know whether it's it's new organizations or other new organizations like fathom I'm sure having all these sorts of discussions with different parts of the financial and corporate sector well wouldn't it be exciting if as well as that we could through groups like yours we could pull we could we could knit things together to create an integrated financial system that dealt with these risks effectively and make sure each cottage industry in the city of London was doing its bit to manage and share it because ultimately this is a governance issue how do we share and manage risk to have a sustainable future okay we'll move on to the next question so I think it's you go next and it's a question brother okay sorry but you said that our exposure was evolving in emerging markets with land use change and in that arena I think insurers are facing what they call the protection gap meaning that people and locations be it California or Pakistan can't afford the insurance but the risk is transmitted down the supply chain into the corporates who depend on for instance Pakistan for textiles and rice and other commodities isn't it time for a bigger picture view of applying insurance climate analytics with TNFD on a whole country basis for instance issuing sustainability linked green bonds or sovereign bonds that are sustainability linked on a whole country basis so you connect together in the case of Pakistan the cyclone risk coming in one way and the deforestation in the Himalayas silting up the rivers with flooding coming downstream the other way to combine and the risk is with marks and expenses but it's a whole country risk yeah I'll I'll try and give a very quick response it's probably more of a response of a drink in the cocktail reception but I will take slight issue the reasons why we have massive under insurance in developing world as well as quite considerable under insurance in the developed world is not just about affordability there's all sorts of reasons why we have low insurance penetration around the world affordability is is important but as I said most insurance is bought because of public policy requirements you buy it because you have to and in some countries it's become explicit in other countries it's not so explicit but your but your general point is absolutely correct this risk exists it is being held by someone usually the risk is being held by individuals who can't cope with it or the sovereign and we have to make this risk and that's surely a a raison d'etre of the cgfi make implicit risk explicit see who owns it and and understand that we might need to develop a much more radical view of sharing risk within countries and between countries to actually manage share and ultimately hopefully reduce it and we need a risk sharing revolution just as we had in this country in 1946 1945 or uh to over 100 years ago in 1906 i 2007 we have completely lost the idea of insurance being an institution of society we think about it as a rather grim grotty product it is not it is actually how society shares and manages risk and we are not going to confront this issue until we become much more focused and explicit i was very excited when the shadow chancellor talked in a speech in in the us a couple of weeks ago about secure anomics i don't quite know what it means but i like the sound of it we have to focus on creating a consistent system of making ourselves more secure and clea against these risks so we can then actually have the confidence to grow i don't want to undercut your point because i think this is very important i think the protection gap largely exists in residential lines of business i mean commercial lines of business are largely insured across the world even in the developed world so just to be aware of that is something that might get in the way of the of the project i'm sorry we i'm kind of very aware that we're running out of time i'm kind of you know don't want to keep you from your drinks i'm not going to take any more questions from the floor i'm just going to ask for final remarks from the panelists i guess we've we've talked a lot about you know kind of sharing data and understanding uncertainty we all have our different perspectives we kind of want to get out of so my final question is how do we how do we keep talking to each other how do we bridge the gaps in translation between what we're doing in research and what we're doing in finance or even different parts of finance so maybe we can start with with Jen to begin with well i guess events like this are a key key way to to get us talking to each other and understanding the needs of different groups um i think yeah sharing data um so that we can really understand as Hannah said before like what events what compound events that we're looking at actually had impact um so we can need need that information in order to um better understand the the compound events um so yeah keep having events like this where we can talk to each other and share and share our knowledge and find out what each other needs really yeah i totally agree i think i'd add that if you see a quarter industry i guess if you see things that we're doing that look nearly useful we are quite flexible like we we like to make pretty pictures to put in the papers but like in theory the the methods we use everything is very adaptable you know and all these papers have email addresses on them um so you can contact us and be like oh this would this if you just change this a little bit this could be really helpful to us we could use it that's actually helpful to us because we need to demonstrate impact to our university so like literally everyone wins um but i think just don't be afraid to ask if things look potentially useful but not they're not quite in the right form which i guess is often the case definitely agree yeah i've usually found when i write to authors of papers they're very helpful yeah to talk about the research yeah i guess for me i think i think we need a professional class of people who are interpreters um in between the financial sector and and insurance and the academics and i guess i'd like to think of myself as one of those people and i think there are a group of us well i definitely think there needs to be more there's more of us now in insurance and i see increasingly ads for banks and investment firms for these kind of professionals and i think we we need that because it is a skill i think to interpret between the two camps get involved in practical exciting projects together i mean that's where it really hits the road when you get involved in practical exciting maybe even commercially relevant projects there's some great ones happening at cgfi like obviously the loss that the wind and flood global resilience index unfortunately nicola couldn't be with us today but that's a pretty exciting project that's trying to cut through some of the sort of challenges we've just discussed but that's when it gets real that's when it gets exciting thank you very much for finishing on that positive no i'd just like to thank everybody for all your questions and your time today and finally we thank the the panelists for their fantastic presentations and comments so i think the next item of the agenda is closing remarks closing remarks it says 15 minutes but don't worry it'll be it's it's only half an hour no it's 15 seconds 15 seconds big thanks incidentally also to Len but also to Paul cb frs for pulling this panel together thank you so much for that right well what a day it's been fantastic and the grandfather was about to hand things back to Ben Coldercott but i mean whether it's baronist pen this morning obviously seabares the talks we had on the transition task force obviously david and nature those of you went downstairs at that at that bifurcation we had some absolutely such exciting new teams and startups which represent i think we have eight presentations from companies that cover quite the range of activities we've been talking about today so thank you all for showing us your great work and i'm sure there'll be a lot of people coming to talk to you in drinks of course finishing with floods so thanks to everybody for staying to the end and we'll have some drinks in a second i do have some thanks to give apart from of course to to Ben uh uh uh the session leads right from the morning uh jack morris uh mike wilkins uh hymen at alvarys paul of course on len uh ian clatcher and ashley sladen can i thank all of our session leads plus all the wider colleagues at nerk and innovate uk who helped make today happen and of course also underpin everything that's being done at cgfi and serif uh i want to make a particular thanks to the cgfi uh central team so hymana alvarys again uh christof christian uh yannach padeher alex horgan and alex jackman thank you all so much and that's been so wonderfully we're in here to anyone else that we've missed so thank you to all of you and let's go have a drink