Hurricane season has been an insurance touchpoint for decades, but development patterns and a changing climate are exposing insurers' books to a wider range of perils. This forum will explore how carriers can use digital to:
- Understand the risk profile of their books of business
- Proactively reach out to customers to ensure they are adequately insured
- Respond to claims faster and more efficiently
Transcripts:
Transcripts are generated using a combination of speech recognition software and human transcribers, and may contain errors. Please check the corresponding audio for the authoritative record.
Nathan Golia (00:09):
And I am here with is Yasir Kaheil
Yasir Kaheil (00:11):
Yes, Kaheil.
Nathan Golia (00:12):
Yasir Kaheil. Sorry about that. I had practiced it in my head during the entire introductory video and just booted it at the end as is my want. So anyway, thank you for coming in and talking a little bit about some insurance digital strategies for changing climate risks. Before we get started, I did want to talk a little bit about why we're talking about this subject and what we can expect today. We have been talking a lot at digital insurance and you've seen our coverage going back to August on how climate change is impacting, how insurers are going to be doing business going forward. And especially with the changing nature of understanding around coastal communities or communities in other sort of dynamic areas. I think we're going to focus more on areas adjacent to water today, but the communities everywhere and their development patterns that are bringing them into areas of dynamic weather activity. At the same time that weather activity is becoming more dynamic because of climate change. There's a lot of new data, a lot of fast moving data, new assumptions that are coming in. And I just got to start by talking a little about what happened last night where I live in central Texas.
(01:31):
We actually are in the process of relocating. So we're leaving central Texas and of course we have sold our house and we're waiting out the last couple of weeks before closing. And last night there was a tornado warning very close to our house within about, I would say 10 miles of our house. But this tornado warning was coming with a squall line of thunderstorms with winds up to a hundred miles an hour, even if they weren't rotating in some spots. And definitely where we are definitely over 60, 70 miles an hour. And of course we're sweating it because the last thing we need when we're about to close on our house is the roof to blow off. But we were fortunate that we did not get heavily impacted by the system, but however, on the weather map as we were watching it coming in on our local TV station here, they were referring to neighborhoods in the town.
(02:32):
We live in central Texas that have names because they are big neighborhoods, big developments that are named most notably Sun City and Sarata. If you're familiar with the central Texas area, these are huge housing tracks with hundreds of houses, people relocating there. And as recently as 20 years ago, they were probably just empty ranch fields. And so when storms like this came through there, house after house, after house, after car, after car after car, parked in the direct path of lines of storms like this. And that's the kind of thing that I'm thinking about a lot when I'm reading about the climate and its impact on insurance is it's not just about the climates changing and everything else is staying the same. We are talking about patterns of development that insurance companies are interacting with and that they're ensuring the construction and the homes that are pushing out into areas of dynamic weather activity. We just had. And we'll talk about a little about Hurricane Ian in Florida, which I think even highlights that a little bit more. But first I wanted to get yes just a little bit for you to talk about your journey to your current role as VP of PNC Analytics for Swiss Re. Can you tell us a little bit about your background?
Yasir Kaheil (03:52):
Hi Nathan and hi everyone. My name is Jessica Hill and I work currently as a vice president of the lead property analytics in the Americas region for Swiss three. And my journey to insurance and where I am today started in 2008 when I started my postdoc after doing my PhD in civil engineering focusing on hydrology, I did my postdoc at Columbia University, which more specifically at the International Research Institute for Climate and Society. And over there we worked on an index or parametric insurance program in several countries, actually like three or four countries in Africa. It was mainly for crop yield insurance and it actually had Swiss three as the sponsor or the carrier of this program. After that I worked in Canada in the weather center on rainfall analysis and then I went this time officially to insurance work for RMS on CAT modeling, specifically on generating the rainfall synthetic set for the US flood model, which was released a few years ago. And then I worked at FM Global, which is a property insurer for commercial and industrial properties. And now I'm basically in my third year at Swiss three in property analytics, property insurance analytics. So it took a few steps moving across the country, east to west and then east and now I'm still in the east coast. Yep.
Nathan Golia (05:53):
Can you tell us a little bit then about how you've seen the kinds of data and modeling that you're working with become? How has it been useful to insurers? What are some of the things, I mean you started out talking about that you were working on this parametric program that was Swiss Re was the sponsor of it, but it was something external to the insurance industry. We're going to talk a little bit about parametric later, but can you just talk a little about how the data you work with is used by the insurance companies you've worked with and worked for? What are companies looking at?
Yasir Kaheil (06:24):
So over the last 10 years, there has been a big revolution in data science. I mean actually the whole term of data science maybe developed 10 years ago or something. And in the past people didn't, less trained people I would say didn't have access or didn't know what to do with things like satellite images or rainfall data or weather data and how to leverage that to basically have more accurate assessment of losses and especially on the cat side in insurance. So in the past, insurance companies just used regular statistics like they do with any type of casualty insurance or Kaheilth insurance or life insurance. So they would do more on the claim analysis and try to figure out what losses they were dealing with and the cohorts and how they can do the rating. Over time, obviously people developed more understanding and more feel for this type of data.
(07:35):
Now they can utilize rainfall data, temperature data. They started using modeling, see how a flood model can help how observation of storm tracks or tornadoes, severe weather, atmospheric models, hurricane models. And they started developing more understanding and more utility of this type of data. And that all started with cat modeling companies like RMS and A IR and these companies that basically put the science in the insurance industry and that kept developing. Now that we have a huge chunk of the operation and insurance is data science is basically processing data and trying to make sense of what we have and how we can use it to compete better as an insurance company, how we can have more access and more accurate assessment of our loss.
Nathan Golia (08:42):
Would you say it's become
Yasir Kaheil (08:43):
More premiums accordingly
Nathan Golia (08:46):
When you say it's become more of a focus on predictive with data science than reactive through the traditional way? You mentioned looking at claims data almost exclusively, but maybe you could talk about the switch from reactive to proactive or predictive I should say.
Yasir Kaheil (09:00):
That's a really good point. I mean risk management in general used to study the history. I mean used to be more like let's say what happened in the last few years and try to make sense of that and assume that this pattern or this behavior is going to persist for the next year and this is where we set our rates and this is how we access the market. But now it is more of like it took that leap from passive to active. Now it is active risk management. What are we anticipating, what the data is pointing us to, how things are evolving and let's deal with that as the precursor for next year. So this has become the nature of the work now that everybody wants to know really what is this leading to, not what has happened and how we can kind of continue what happened now we want to know what's going to happen. That's the difference That basically leads to the big topic of climate change that everybody's now looking at. Sorry, I interrupted you.
Nathan Golia (10:10):
No, no, no, it's okay. I know that sometimes these web webcasts can be a little, you get across talk and it happens. But when we did our pre-call for this, you said so many interesting things about just little things that I sort of instinctively could understand but did not realize that there was actually data to back these up. You're a hydrologist by trade, is that correct?
Yasir Kaheil (10:29):
Yes, I am.
Nathan Golia (10:31):
And you mentioned something about how storm activity works that storms are changing. Could you talk a little about how a storm has changed over the time that you've studied hydrology or maybe even using some of the historical data to how things are working now?
Yasir Kaheil (10:49):
Hydrology, I mean basically just studies how when the rainfall or when the water interacts with the surface of the earth and how runoff happens and how reverse form and how much water there is in the river and how fast it is and so on. But the storm starts from the atmosphere. I mean what's happening now, and this is one of the hypotheses that people say because of the warming and the atmosphere, now the atmosphere has more capacity to carry water. And that means that instead of having, once the atmosphere is saturated, it drains, but now the atmosphere can carry more water. So the frequency or the number of storms that you will get is fewer, but they themselves, the storm itself is bigger and more intense. Now intensity is extremely important. So when you get higher intensity in each storm, first of all the soil has an infiltration capacity.
(12:00):
The soil is not going to infiltrate all the water. At any rate, there is a certain rate of rainfall beyond which the soil really cannot absorb the water. So that water is going to start to run off and cause flooding or flash flooding on the way to the river and then the rivers will swell and will not be able to carry that much water all of a sudden like that. So this causes more flooding, it causes more intense storms, the wind is stronger, the rainfall is stronger, it causes all sorts of natural catastrophes and it makes it more difficult to deal with in terms of infrastructure. And then I mean the last thing is the financial hedging of this or of the risk, which is insurance in this case. So now it becomes is our infrastructure even ready for something like this and for how long it's going to be enough? And then whether this financial cover that we always try to go to as a solution, the insurance, is it going to be sufficient? Because at some point if your infrastructure is not absorbing most of the risk, I mean insurance is just for the residual risk. If the infrastructure is not taking most of the risk, then insurance is going to fail too because it's not going to, insurance has limits and that would leave a lot of people being in harm's way. I mean insurance and infrastructure are not helping and who knows, so
(13:51):
It's not good for the future. But what's good is that now we understand the data better. So now at least we have that we can plan for it hopefully ahead of time before it happens. We can evacuate the cities, we can just like what happened recently with hurricane, so we can anticipate it better at least.
Nathan Golia (14:13):
And I think that we want to talk about hurricane in a second, but just to close out on that point, the thing that was interesting to me when we talked about this before was that direct link between, hey, if you're warming the atmosphere, there's more capacity for it to hold water. That means that you may have fewer storms that are going to be more intense. And that got to some other things that I had heard in the news so-called wet bulb temperatures that were going to be increased because of the increased capacity of the atmosphere to hold water. And like, oh, when you see all these things starting to come together, you really do see that climate change is, unfortunately there's a lot of noise around it for certain reasons. But I think the fact that if we're saying the data like hey, this is going to happen and there's really not whether what we can talk about it later, but if it's happening and these are the cases that were happening, the insurance industry is going to have to catch up to understanding that risk.
(15:05):
And to me, I think there were a couple of things that happened. Again, one close to home here in Texas earlier this year, it might've even been right before we spoke in Dallas, there was a lot of flooding because of very intense heavy rainstorms after as again you talk about the soil, we had had an incredible drought in Texas for much of the summer, even though it was still humid. We hadn't had rain, but it was humid and damp in the air. And I guess I sort of saw in real time over the course of this summer the way that the atmosphere was holding more moisture and then it kind of came out all at once in a big storming event. And there was a lot of talk originally about, well look, here's the problem. Yes, all this water is welcome and it's not bad, but we can't harness it.
(15:49):
That gets to your point about infrastructure, being able to understand how to control this water as it's coming down and whether or not the infrastructure cells remove it. And then of course the very unfortunate floods that happened in northern Kentucky earlier this year, again flat, you're talking about flash flooding that was impacted with the environmental choices that had been made around mining up there. And I think seeing all that and then being in the insurance industry, like you said, Hey, insurance companies need to know this is going to happen and that's where the data comes in. Just talking about trying to make this case, Hey, this is what's happening. Here are the new conditions under which you're operating under and here's what we understand. Maybe you could just tie that a little bit together for us.
Yasir Kaheil (16:30):
Correct, yeah, I mean you mentioned another point, but maybe I'll let you get to your question, second question first, but you mentioned something about the pattern of urbanization, how people now are building in areas that weren't inhabited before and that's a huge problem for insurance, especially for wildfire. I know that you're in Texas and you talked about tornadoes and even by the way for tornadoes, sorry, I don't want to go on a tangent, maybe I should let
Nathan Golia (17:01):
You, it's okay. You go on as many tangents as you'd like. You're the expert. I'm just saying, Hey, what is this thing that's happening to me?
Yasir Kaheil (17:11):
So even in the data, by the way, to study severe convective storms like hail and tornadoes, traditionally this data used to be collected by storm chasers and they called them tornado chasers and storm spotters and people like this who are in the atmospheric physics departments and they love to go and chase tornadoes. So they obviously would not go and collect data in an area that nobody reported anything. So the hail data, I think before 90 was not good because it just shows you data where there is a road or where there is a highway or something like this or where there is towns, but elsewhere it shows that there is no hail. But that's not the case. I mean, hail used to fall everywhere, but it happens that nobody was around to report it. So then Noah I think introduced a new program to collect the data and now all of the models are based on a new phase of that data that finally these gaps that we used to think there is no hail in this area. Actually there was, but because we just didn't record it. So things like this also play a big role in modeling these events and understanding. And now where people live now, they are more at risk that in the old world of insurance they appear not to be at risk, but they have been always right, and it's just because nobody lived in that area.
Nathan Golia (18:53):
I mean, you're right, I do live in Texas now and I'm thinking about tornadoes and storms. I used to live at Utah where wildfire was a much more profound risk and definitely have talked multiple times in our various events about looking out from the back porch of my house and looking up at wildfires that were happening on mountain sides where they were also, if you looked a little bit to one side or the other, they'd be building a neighborhood up into this mountain to, because there was no more room in the valley. And I was saying like, geez. And we saw that even we actually did have a minor fire in Texas that was in one of these neighborhoods that again used to only be ranch land within a couple decades, even maybe a decade that where they were pulling water out of our reservoirs to try and quench it before it hit any houses because we're getting so close to a house line.
(19:44):
And yeah, I think that's an interesting, there was a band, they're not notable, but it was a local band when I was growing up, they had a song and the first line was, is it a blessing or a curse to see through unclouded eyes? And when I think about insurance data, that is the question. It's like, hey, it's great that we now know all this is happening, but we're also sort of cursed by understanding it and understanding that these are risks that we have to account for, that if we had gone by the old data and people were just building out there, we would've been sort of blissfully unaware until the event happened. But now we know it could happen. And that's where all this modeling is coming in. Hurricane
Yasir Kaheil (20:26):
Also,
Nathan Golia (20:28):
No, go ahead point then we'll move on to hurricane
Yasir Kaheil (20:31):
One thing, I mean, before we get to hurricane air, I mean a lot of things happened in insurance after Hurricane Katrina in 2005 and the levies in New Orleans and all that problem. And that triggered that study about all of these things that are meant to combat storms and hurricanes and storm surge, which are traditionally built by the Army Corps of Engineers, especially if it belongs to the country or to the government.
(21:10):
And there was a process for the Army Corps of Engineers to build, say, to build the levy or to build a dam, and they say this is going to be enough for the hundred years, but where they came up with that height, that storm height was just somebody's assumption. So they say, we are going to build this to be a hundred feet or something like this, and just somebody came up with that number and now it's a hundred year return period. So after Hurricane Katrina, I mean there was major studies like that where they looked into these structures that people thought were protected, but now they realize that, oh no, this is just somebody called that a hundred year, but in fact it's not. So the storm is probably going, if there is a storm, God forbidden in some areas in the country. I mean it's going to be a big problem.
Nathan Golia (22:14):
What I wanted to talk about with Hurricane Ian was the difference between storm surge water and rainwater. Because this storm was, I remember thinking about it in the week leading up to it. I could not believe it seemed like it was going to be huge, mostly because I love reading all these modeling things even I'll always understand them. I do love reading them because it's so interesting to me to see here's what we can expect. And later following it, especially my wife filing on social media, some people who were living down there and I was saying, yeah, I kind of knew this was going to happen. I'm sort of surprised they didn't expect it. But I'm just curious about your, and especially we're a couple weeks removed from it now we have a storm surge water event, we have a rainwater event. How do you as a modeler look at those events and say, here's what we can learn from this storm and put into our models for insurance companies to use.
Yasir Kaheil (23:07):
I mean, storm surge water comes from the ocean, right? I mean it's basically the wind pushing the ocean to land. And Hurricane Ian, I mean something very interesting happened because the eye of that hurricane was between Tampa and Fort Myers, I think, and the hurricane was going counterclockwise, so it was pushing water onto Fort Myers, but sucking gu out of Tampa. So they had negative stone surge. In Tampa, you can actually see the sludge and the bed of the bay, whereas in Fort Myers, they're dealing with tens of very, very high storm surge as
Nathan Golia (23:52):
High as 12 to 15 feet. I think it,
Yasir Kaheil (23:56):
Yeah. So that's storm stone surges. It's basically wind pushing ocean water onto the land.
(24:03):
Now after this, this is of course also a big storm. I mean there is clouds and these clouds are coming inland and eventually it moves, the entire storm moves inland and brings with it all these clouds and all that rain. So that rain, it's going to precipitate and it's going to fall at some point. And like I said, these are wet conditions. The soil is already saturated so when the rain falls, it's going to cause flood. But in coastal cities, luckily, I mean it drains quickly into the ocean. Now the problem comes is after the storm moves more inland and you have all the strain coming onto these areas and it is intense rain, it's definitely going to cause some kind of flood, either flash flood or it'll fill the rivers and the rivers will flood later on. So the way to model these two things are very different. I mean, one is the wind modeling and the storm surge model, which follows the bathymetry of the ocean and the ground elevation. And then the flood modeling, the inland flood modeling is just hydrology and hydrodynamics. So these are different sciences even.
Nathan Golia (25:27):
So
Yasir Kaheil (25:28):
The way this is being handled in sciences, like two separate processes, the inland flood model, by the way, in different cat modeling companies, they distinguish between inland, sorry, precipitation caused by a hurricane or by a tropical storm or tropical or tropical cyclone and just regular precipitation that is from a winter storm or from regular thunderstorm and stuff. So these events of the hurricane induced rainfall, this is treated differently and goes into the synthetic set of the cat model also in a different way added onto the synthetic set. It's not part of the regular rainfall. They remove it even from the inland flood because also in insurance it is treated differently because it's under a named storm or under an event. So yeah, that's the difference between the two.
Nathan Golia (26:35):
I wanted to just wrap up here quickly by talking a little about your experience doing parametric insurance. It's something that we actually have, as recently as yesterday, I had asked one of our other editors to put together a list of some of the parametric programs that are rolling out. We are seeing this as a sort of new way or an emerging way at scale to attempt to close some coverage gaps. We, we've got not just our ability to understand risk, but also our ability to see when it happens and trigger payments without having to do a lot of claims, without having to do a particular claims process. And I'm just wondering about, do you think that these kinds of programs are something that we'll see more of in the future to bridge the gaps between not having any insurance at all, which unfortunately we found out after hurricane, a lot of people were uninsured and the expense of insurance that it takes, just trying to make sure that people have some cover. What about the ability to use data makes parametric insurance more scalable now? Why is it being able to be used more?
Yasir Kaheil (27:50):
It's more scalable because of what you said. I mean, it's more efficient as an operation, right? I mean, there is no claims adjustment. There is no just if the index is triggered or the parameter that you're measuring is triggered, then there is a payout and that's it. The characteristics of that index or that variable that you're tracking should be that it is known to everybody. It is easy to measure, it's known to everybody. And it's also, you can't tamper with it as an insured entity. So it makes much more efficient as an operation. The only problem with it is that sometimes it doesn't correlate so much to the loss. I mean, sometimes the trigger, the index is triggered, the payout is triggered and there is no loss. And sometimes the opposite happens, which obviously is much worse for the insured, the entity that you have a loss, but your insurance company says, sorry, we don't have indexes is not triggered. So it's not good in that case. But as scalable, like you said, as something that is scalable, if you ensure the trigger, the index is correlated with the losses, I think it could, has been a solution to a lot of these problems, especially for government entities, I think, where they don't want to do a lot of the structure angle of the program itself.
Nathan Golia (29:34):
And I think that maybe you're a data guy and that other question is for the front end guys to figure out how to make sure that they're matching everything correctly. But I guess from a data infrastructure perspective though, these things are so accessible that it's really a matter of program design from there.
Yasir Kaheil (29:52):
Yes, yes. Yeah, absolutely.
Nathan Golia (29:55):
Ya. Thank you so much for joining us today. It's been about half an hour. We're going to close it out there and we'll follow up. Feel free to follow up and keep an eye on digital insurance for developments out of Swiss Re.
Yasir Kaheil (30:10):
Thank you, Nathan. Thank you so much.
