Hi, I'm Gabriel Cohen, host of the Verisk podcast series Verisk Talks Risk. We hope you had a chance to listen to part one of this podcast, Preparing for the Impacts of Climate Change, where I spoke to Ed Crooks, vice-chair of Americas and Amy Bowe, head of carbon research, who both work at Verisk’s Wood Mackenzie business.
Ed and Amy shared their insights on the implications of climate risk for business and our physical environment, the costs of curbing greenhouse gas emissions, and ways to reduce climate risk and more.
Today in Part 2, we’ll be joined by Shane Latchman. Shane is vice president and managing director of Verisk’s AIR London office, and he's an expert on the impact of climate change on business decisions. Shane and I will discuss some of the issues around how the insurance industry is responding to climate change risk and how insurers view the future, in terms of priorities and the changing landscape.
Well, welcome, Shane. It’s great to have you on and why don't we get straight into it and start with asking you how is the insurance industry evolving with regards to modeling and mapping climate change risk?
Thanks for the question, Gabe, and great to be here.
I think all of this really is quite nascent; this use of natural catastrophes and the quantification of climate change risk in general. Natural catastrophe modeling has been used really to quantify current risk. So for insurance companies looking at what's the risk to my policies over the next year; for example in 2021.
And it's only recently I'd say within the last couple of years or more that natural catastrophe models and other tools have started to be used to quantify the risk due to climate change in the future. Largely, I would say the motivation for that has been regulation.
In 2019 here in the UK, the Prudential Regulatory Authority, which is part of the Bank of England, announced their general insurance stress tests and as part of that, firms had to quantify their risk of climate change and how that could change the losses from an average annual loss point of view, which is the amount you'd expect to lose every year on average, and one in 100-year loss for a range of region perils across the US. For example, the US hurricane. And here in the UK, for example, freeze and subsidence, and rainfall and sea-level rise.
And that really kickstarted a lot of interest in quantifying climate change risk. In this year, 2021, the Bank of England has their CBES scenarios, which is their Climate Biennial Exploratory Scenarios (CBES), and so the industry will need to quantify changes in climate risk for a range of natural catastrophes, and also other types of risk, not just on the physical risk side but also on the liability and transition risk.
And the way that that's done is through a set of tools. For example, what you could do is you could take your existing catastrophe model and you can identify trends in the climate. You could say, for example, that sea levels will rise by X inches, and you can put that into the model and you can get a new view of how catastrophes will be in the future. Or, for example, you can say that storms will be more poleward and hence go more North in the Northern Hemisphere and as a result, you can then put this into your catastrophe model and you can simulate how storms would look like in the future. That's on the model side. As to regard your mapping part of the question, what you could also do, for example in the flood hazard map side of things, is look at the impact of climate change on flood hazard maps. So the extent of rivers and flooding at the coast as well can also be done.
That's really interesting Shane. I wanted to ask you a two-part question now.
So given the perceived escalation in climate change risks, how true is that we can still use historic data to predict the future?
And if we can't use historic data, how is the market using data more intelligently to quantify these risks?
So if you look at the media, it would seem that every catastrophe that happens these days, that's a hydro-meteorological catastrophe so flooding or a storm—you would think that they are all connected to climate change and the casual reader or viewer may think that climate change caused this specific event.
But of course, first things first, climate change might make some events more likely, but won't necessarily cause that event in itself. The event might have happened anyway, and I think part of climate change is then thinking that if things are changing so much in the future as to your question, then what is the use of historical data?
I think the specifics here are really important. Climate change, we know and we have confidence is impacting sea levels. Sea levels have been rising and will continue to rise. We also have a lot of confidence in knowing that climate change is impacting certain perils like wildfire, flooding, droughts.
However, we have less confidence that climate change has already impacted other types of events. For example, tropical cyclones and so on.
And so hence, when we're looking at building a view of the present, so the near present climate risk, we can definitely use historical data to calibrate and build our models to represent the current risk. And if we're looking at risks over a time horizon of a few years or perhaps a couple decades, then the historical data provides a lot of good insights into building models and representing risk in the future.
One fact that is not widely realized is that when you build catastrophe models to quantify the risk of hydro-meteorological perils—windstorms, floods, and so on—the length of the historical data set is actually only about four decades in general terms; there are few exceptions, for example, U.S. hurricane. But they're not on the orders of hundreds or thousands of years because earthquakes, for example, have longer historical datasets.
But when we think about hydrometeorological perils, we're really just looking at the last 40 years or so. And hence when we build catastrophe models to some extent of course, given that they're built on the last few decades of data, they will represent to some extent the current risk and of course can be used then for the future.
The one last part of my response is in terms of further time horizons. So if we're thinking about longer time horizons, so several decades into the future towards the end of the century, I really think that whilst historical data is useful, it probably becomes less useful the further out you go. And this is where we need new types of science, new types of methods to build and quantify risk from climate change. For example, building physical models and estimating what (events) will look like in the future and from those physical models, creating the events themselves that will happen in the future.
It's interesting to think that 40 years in a data set isn't very much, but when you're thinking about major major events like earthquakes and it really does put it in context.
What about the pandemic? So there are so many aspects of the insurance industry that have been disrupted and in many ways trends accelerated due to the last 18 months. How do you think the COVID-19 pandemic has impacted the insurance industry's approach specifically to climate change risk? And how do you think this could influence the future?
I think one interesting artifact that we observed during the lockdown and all the various lockdowns and thereafter is at that point companies were quite reticent to make large changes. They looked at items like nice to haves and said, well, a lot of things are changing. Let's just stick with the status quo. Let's just stick with what we have.
The one exception that I saw to that was, of course, climate change and at some level, the COVID-19 crisis can seem existential and bring to mind other existential crises that we as human beings have been hearing and ignoring recently like climate change.
This concept of being more resilient and so forth that definitely resonates with business, but I do feel apart from the pandemic it's all about timing and the regulatory part of things. Which is that there certainly were regulatory requirements that were happening around the same time. I talked about the 2019 General Insurance stress tests (and CBES) as part of the Prudential Regulatory Authority.
Companies would have been working up towards and now providing and understanding their exposure to get prepared for filling out the (CBES) requirements that start in June and for which they have to report in September. So that's certainly something that, while this pandemic is going on, companies need to be aware of. And also another interesting thing happened, which was in July of 2020. The PRA published a “Dear CEO” letter and in it, they said that firms should have fully embedded their approaches to managing climate-related financial risks by the end of 2021.
And that also goes on to say that capital levels need to adequately cover the risk to which the firm is or might be exposed. So all of that really points to firms needing to and wanting to focus on climate change, but of course, even though there's a pandemic and we're focusing on climate change, I think you want more of it than just regulation. And we do see firms looking at this, not just for submitting regulatory requirements or satisfying those, but also seeing how this can then play into business decisions. How can this impact strategy; writing more for example, in certain areas and less in other areas. If, for example, we know that storms are moving more North and could avoid countries that you know there could be an increased risk in some areas than others, and we're seeing companies trying to embed these in their decision-making much more.
So if I'm an insurer, I have to place some sense of prioritization on what I'm really going to focus on when it comes to my climate risk priorities. If you were advising me, which are the natural hazards that you think that over the next 10 years as an insurer, I should most be prioritizing when it comes to climate risk?
So within regulation within Europe there is the regulation of Solvency II as many of you will know and with Solvency II, there's this wonderful principle called the principle of materiality and proportionality, and it's very useful for regulatory purposes, but also in life. And you probably think that quite funny that there is a principle in regulation which sounds not that interesting that could be more widely applicable but let me explain it.
Essentially it says that the risks that are more material you spend proportionately more time on, and you can imagine parallels in your own life where that might be useful. And so within AIR, we have this wonderful output called the global exceedance probability curve. I'll just take a few seconds to explain it. What we do, it's very unique. I think we're the only ones that have this output. We take all of our catastrophe models everywhere in the world, around 100 countries around the world, for many region perils, and we intersect those models with our estimates of exposure. So how much buildings, contents, etc. is of value in these different countries.
When we do that, we can then create an exceedance probability curve for the entire world and within that output, there's a very interesting result, and that is that at the 100-year TVaR—so the one in 100-year 1% tail value at risk—on an insured basis of all the exposure around the world, 60% of it is driven by U.S. hurricane, so U.S. hurricane is driving your overall tail risk.
So if I were to give some advice, I would say for other principles of materiality and proportionality, whilst we may not have as much confidence about U.S. hurricanes as we do for other things like sea-level rise and droughts and so forth, well maybe you should spend a good proportion of time there looking at ways to stress and sensitivity test any changes to U.S. hurricane risk.
So then, apart from that principle, we think about this other principle about confidence. Where is the science confident?
One interesting result we had recently. We just updated our U.S. inland flood model last year and something that we observed is that in the decades closer to the present, the last three decades or so, we found it to be more wet than the decades further towards the middle of the last century. So essentially, if we were using a very long time series, what we're saying is that the time series that is closer to now is more wet than that further in the past, and actually, that goes back to my point about using more recent data and we made some explicit adjustments where we were seeing a trend and essentially made the model that we released more wet than we might have if we just used a stationary assumption of the overall precipitation in that more recent record.
So really, we're focusing on things where we've seen signals, and so, for example, back to your point, U.S. inland flood is something that we will look at and flood models in general. Wildfire models are also something that we're looking at. So when we update our U.S. wildfire model in 2023, we will be looking to see how we can incorporate climate change. What are the signals that impact near present climate, and to what extent we can use the full historical record or whether we need to impose a non-stationarity assumption in there?
I love the idea of applying the principle of materiality and proportionality to our personal life. I mean, it doesn't seem that humans are perfectly rational, but that's probably a discussion for another podcast.
A final question I have for you, Shane.
We've been talking about a lot of trends, and this is certainly a big topic. Do you think that insurers that seem to be doing the most in terms of combating the effects of climate change will have a competitive advantage in the future, especially in the eyes of consumers?
I think that's a hard question to say with great certainty. I mean, it makes me think of Niels Bohr, who once said “prediction is difficult, especially if it's about the future.” I think you know definitely there's some evidence of recruiting within oil and gas, and you would hear from our colleagues at Wood Mackenzie and some of that recruiting has become more difficult recently, given the scrutiny being placed there. This concept of social purpose, being good to the planet, green credentials, and that's especially important for these Gen Z+ millennials.
And I say these Gen Z must plus millennials, but technically somehow I'm a millennial, even though sometimes I find that hard to relate to. But in a big way, I think it's tied to branding; it’s tied to marketing. And if large insurers are marketing green credentials, that could help in terms of consumer purchasing patterns and also attracting talent to work for those firms.
Climate change is also a slower churn. It will likely manifest in nonlinear ways, likely resulting in unprecedented weather extremes as we talked about earlier. And so companies that are best prepared for these outcomes and more resilient, hence will likely have an advantage and also those companies, not just the insurance companies, which begin looking at their decisions like long-term investment, strategy, through a climate lens and begin building a climate aware corporate culture, I think will definitely benefit in the long run. And we see firms nowadays as I mentioned before, increasingly interested not just to do this from a regulatory point of view, but to implement and incorporate this into their business structure, and I think we'll see a lot more of that, and I do hope that the planet will be better for it and I feel optimistic about that.
Yeah, I think that's a great, great note to end on. Nothing like ending on something that relates to the future of humanity. Shane, thanks so much for joining us. This has been a fantastic conversation.
Thanks a lot, Gabe. Great to be here.
To learn more about the risks impacting business, communities, and people, be sure to visit us at Verisk.com and follow Verisk on LinkedIn, Facebook, and Twitter.
We hope you enjoyed this podcast and we invite you to join us for our next podcast on Financial Resiliency in the Payments Ecosystem. Until next time.