Four Thoughts on Climate Change and Catastrophe Risk

January 6, 2020

Climate change. Is there a hotter - pardon the pun - topic in catastrophe risk at the moment? New companies and consultancies are popping up offering advice and guidance on how to cope with this topic that has escalated in boardrooms greatly in the past few years.

 

Here's a few thoughts of mine around the topic on how catastrophe risk might pan out in the next 2, 5, 10 years. They're just opinions, so you're more than welcome to disagree! You'll notice I've put quite a few hyperlinks in here to help read around a few things that are driving my thoughts at the moment.

 

A - Expect shifting views of risk on climate

 

Ever rolled your eyes when the latest USGS view of risk leads to seismic (whoops, another pun) changes in the earthquake numbers of your favourite cat model? This is merely what happens when the science is still regularly being updated about the underlying hazard. Also, fairly small changes in hazard are amplified greatly through vulnerability: don't blame your cat model vendor for this. 


Why have I mentioned this? Climate models have been doing a good job for years in predicting how the warming of the earth has changed in the last 15-20 years, which is good news. However, when it comes to understanding catastrophe risk, typically the events we need to model and understand have very sharp hazard gradients that require high resolution to resolve, especially those in the tail. Climate models, with their coarser resolutions, struggle to represent the tail risk well. As computer power continues to improve, we will probably see tail risk better represented either through better model resolution, or better methods of using climate models for understanding tail risk.

 

Recent pieces of work on climate modelling for European Windstorm and US Hurricane (the latter is yet to be published at the time of writing) have both highlighted how the climate response to a model's resolution and its other model configurations can be quite variable as we look to the future. I expect more of the same as climate model resolution continues to improve and more money is thrown at the climate models we use to tie down future (and present-day) climate impacts - if any.

 

We are still learning lots about climate change, so expect our understanding - and numbers - to move as we understand the topic more and our models improve to represent it.

 

B - The relevance and utility of historical data in catastrophe risk

 

I think we're at a tipping point of how we use historical data. This one really bothers me, on three counts. I've written about this topic before on this blog.

 

Firstly: if climate impacts mean the past is of no relevance or a poor representation of the present day, why are we using historical data to understand risk? Should we be couching our risk in present-day data and placing less stock on data in the past? Naturally, this means we have more uncertainty as we shorten the historical record, but this is where we need a re-think about using other data sources (e.g. climate model data) potentially to create our present-day landscape of risk. If risk isn't static, historical data is yesterday's news.

 

Secondly: our historical records are simply too short to understand trends in historical data in any case. If we have 120 years of, say, HURDAT data, how are we to use this to understand how the tail risk is changing? Up until recently we've been comfortable with a static view of risk, but if climate change is responsible in some way for forcing rates later in the 20th and early 21st century, we need to understand how this is impacting tail risk and climate models, for all their foibles, could be the way forward. 

 

Thirdly: maybe less related to climate data, but still on this topic is how representative is "our" history of the underlying risk? Some work I've done on Europe wind has shown that re-simulations of the last 60 years of European Windstorms can be wildly variable depending upon which simulation we look at. There is no other ground-truth than history, but how reliable is it a representative of the underlying risk?

 

I worry that relying on historical data to understand present day risk levels will leave us stuck in the past.

 

C - Listen to contrary opinions: don't shut them down

 

A while back on social media, I was taken to task for pointing out that the US Hurricane drought (the lack of Cat 3 hurricane landfalls in the US from 2006-2016) was never spoken about in the same breath as climate change. It was pointed out to me we'd just had two active hurricane seasons at the time of making this comment, so why on earth should I be linking the drought to lack of hurricane activity?

 

I felt my thoughts were reasonable and worthwhile: some future climate model output has hinted at a shift away from US landfalls in the future, so this period of lower landfalling activity could have been an imprint of hurricane behaviour in a warming climate (see the chart below that highlights the change in future tropical cyclones from 2051-2111 from the aforementioned paper).

 

 

 

My concern here is that "climate change is making all catastrophe risk worse" is now almost dogma: we (and that includes me, at times) subconsciously look for worsening impacts when looking at things through a climate change lens. We're partly spoon-fed by the media selling newspapers, getting website clicks and more TV viewers: the decreasing frequency of catastrophes doesn't make good TV. 

 

Quite simply, we have to remember that first and foremost as an industry we want to understand the risk, whether climate change is making things worse or not. I convened a panel for the ISCM in London last year and the panellists - all Meteorology PhDs - expressed a concern that taking a contrary view on climate change would have you labelled in the "denier" camp when really we're just trying to understand the evolving risk.

 

For example, my current position on European Windstorm is that the magnitude of changes in loss during the warming in the last half-century are less important than other changes brought about my Europe's population changing (which I'll detail in a blog post soon). I would hate to think I could be labelled a "climate denier" for having this view. However, I'm very willing to change this view given alternative evidence.

 

Open, constructive debate is key in the climate change / catastrophe risk debate. Let's not put people into "believer" or "denier" boxes and always be prepared to change your stance given it's such a relatively young science.

 

D - Climate change and "choice"

 

I see Climate Change becoming the modern-day "Near-Term View of Hurricane Risk". Even now some people advocate the long-term view, others the near-term or warm-sea view of landfalling rates (often depending on which end of the insurance transaction you're on!)

 

There will be different views on how climate change is impacting present-day and future risk offered to us in the future. Given the uncertainty around climate and the immaturity of the science of climate change and tail risk, developing your company's view of climate for each peril is something that will come to the fore: if you're not already doing it.

 

Whether this view comes from

  • vendor model alternative rate sets 

  • views of risk developed from companies already nimble in the climate space (for example, see JBA's work on future impact of climate change on flood risk)

  • use of academic data or studies supervised by your company

  • in-house work from the scientists you employ now or in the future

...to help answer this internally, it will be your choice and I'm sure different views will result.

 

I expect many more panels at cat modelling conferences on views of climate risk, quite possibly discussing contrary opinions!

 

 

Please reload

Featured Posts

A very brief hello! I intend to use this blog to write on anything from a topical subject to a distillation of research that will appear on the...

A quick blog introduction

September 18, 2016

1/1
Please reload

Recent Posts
Please reload

Archive
Please reload