Science lesson: Tropical cyclone risk and exposure in a changing world

Authored by Cat Pigott, AXA XL, Head of Science and Natural Perils, Reinsurance

Climate change represents an unprecedented challenge for business and societies and the (re)insurance community to assess, manage and transfer the risk related to natural perils. Cat Pigott, Head of Science and Natural Perils at AXA XL, discusses how working with scientists to understand the changing profile of certain perils and exposure to them will help underwriters and risk managers to better understand the way climate change may lead to various potential loss scenarios.

Climate change and its effects represents one of the biggest challenges facing businesses and societies across the world. Understanding not only the impact of climate change on severe weather events, but also how exposure to those events will change in future, will be vital in helping the (re)insurance industry to work with clients to assess, manage and transfer their risk in the near, medium, and longer-term future.

Engaging with the scientific community to better understand and model risks and examine future loss scenarios is an important part of our work here at AXA XL, as we try to grapple with the effects of a changing climate on our clients’ risks and our own exposures.

We recently held a science-based seminar to explore the potential effects of climate change on tropical cyclones, identify the drivers of this changing risk and the best ways to refine climate-change adjustments to our tools, as well as explore how other risk components, such as exposure – are changing due to urbanisation and societal influence.

In collaboration with scientific partners, we are trying to examine in greater depth how the peril of tropical cyclone and, in particular, the sub-peril of extreme rainfall is currently captured by our risk assessment tools, and how likely it is that this peril will change as the climate changes.

Integrating scientific findings into our catastrophe models and underwriting processes is not an easy task and we have begun a project to help us refine our work on climate-change-related risk adjustments. We also wanted to try to better understand how exposure, notably in areas of urban development, might evolve in the future and whether exposure or climate change is the main driver of change in the past, as well as in the future.

Tropical cyclones getting wetter

Tropical cyclones – variously known as hurricanes, typhoons, tropical storms or cyclones, depending on their location and strength – draw energy from water sources and are, therefore, typically at their strongest when over water. Though they begin to lose intensity when they make landfall, tropical cyclones can still lead to significant damage over land because of the destructive power of their winds and the significant rainfall and resultant floods that they can cause.

Recent tropical cyclone events have highlighted the huge destruction and cost that this precipitation can cause. Hurricane Harvey in 2017, for example, dumped a record-breaking 127 billion tonnes of water on areas of Texas and Louisiana, causing catastrophic flooding that resulted in widespread damage, estimated insured losses of about $3 billion and, tragically, the deaths of more than 100 people. In 2021, Hurricane Ida, the second most damaging hurricane to hit Louisiana after 2005’s Hurricane Katrina catastrophe, not only wreaked havoc in that state but also resulted in tornado outbreaks and widespread flooding as far away as the northeastern United States.

Dr. James Done, a research scientist at the National Center for Atmospheric Research in Colorado and an expert in understanding the physical processes leading to severe weather events, spoke to our seminar audience about the work he is undertaking with the AXA XL Science Team to better understand the relationship between tropical cyclones and extreme rainfall – and how that might develop in future as a result of changing climate. Dr. Done has been using data from 310 tropical cyclones that made landfall or grazed the coast of the United States in five hurricane-prone regions – Texas, the Gulf Coast, the northeast, the southeast and Florida – between 1981 and 2021, to develop a model.

According to Dr Done’s research, stronger, slower and larger storms tend to be wetter. In addition, he has investigated how future climate might affect Tropical Cyclone-Induced Precipitation (TCIP) by taking 13 historical US storms and simulating them in current and future climate scenarios.

This research showed that the distribution of TCIP will likely mean that storms will be wetter in future. Warmer air can hold more moisture, and stronger storms generate more precipitation Dr Done explained, and so as the climate warms, the precipitation that will result from tropical cyclones increases.

Advancements in climate change work: assessing drivers of risk and refining our implementation into business

The second speaker at the seminar, Dr. Steve Jewson, aims to bring a greater degree of scientific research to the insurance industry, notably by improving methods of catastrophe risk modelling.

These include ways to better understand hurricane and tropical cyclone frequencies, intensities, correlations, uncertainty, landfall, trends and changes under climate change. Dr. Jewson explained that changes in hurricane frequency and intensity are closely correlated.

Mathieu Boudreault, Professor of actuarial science at Université du Québec à Montréal (UQAM), and Fellow of the Society of Actuaries and of the Canadian Institute of Actuaries, also spoke at our seminar and presented some of the recent findings from his current collaboration with AXA XL.

UQAM’s Department of Actuarial Science has so far explored the potential for portfolio diversification, while developing statistical and AI / machine learning models for floods and tropical cyclones on a global scale.

In this work, he uses tools previously built as the result of work with us but with a renewed focus on the response of tropical cyclones to various drivers like climate change, changes in exposure and natural variability.

AXA XL and UQAM hope to answer a very important question for our industry: specifically, how much of the change in our risk comes from climate change and how much comes from changes in exposure.

Pilot study on understanding future changes in exposure in California

As we look to the future, it is vital to try to understand the impact that a changing climate will have on risk and loss. But insurers and other risk experts also need to look at the other side of that coin, too – namely, the role of changing exposures on risk. As another of our panellists, Dr. Virginia Iglesias, a research scientist at the Earth Laboratory at the University of Colorado, neatly summed it up: “There are no losses without assets and there is no risk without exposure.”

Dr. Iglesias’ contribution to our seminar outlined the need to understand how changes in exposure might affect future losses, particularly in areas of high urban development. The Colorado team has been studying the distribution of assets within natural hazard hotspots and seeking to understand the effect that urban development might have on losses from extreme weather events and natural hazards.

Five different hazards were considered – earthquake, flood, hurricane, tornado and wildfire – and pixelated on a map of the United States in areas where the probability of them occurring was in the top 10% for the country as a whole. The scientists then calculated the numbers of structures – e.g., such as houses, hospitals, schools and offices that currently fall within each hotspot. This showed that almost 60% of those structures fall within these hazard hotspots. The team also looked at trends in exposure for these hotspots and found that, for example, between 1945 and the present day, development within hurricane-prone areas has been faster than the national rate. Indeed, currently exposure in hurricane-prone areas is three times higher than the US national average.

Dr. Iglesias, in partnership with AXA XL, then explored whether data from these trends could be used to predict what exposure might look like in the future. This is, she noted, not by any means an easy task because of the sheer number of variables, including the physical landscape and the social, economic and political contexts at play in driving development. The vast amount of data now available, however, thanks to the use of social media and artificial intelligence, among other things, means that machine learning techniques and new statistical tools can empower research and modelling in ways that previously would not have been possible, she explained. Using a database containing anonymised data about the location and attributes of structures across the U.S. from 1810 to the present, the team created a model that predicts the future number of structures in California.

While the model, like all models, is not perfect, she said, because it cannot show what might happen if the drivers of development were to change as a result of an event like a financial crisis, for example, it gives us the possibility to begin to simulate future development and understand future potential exposures.

As the world’s climate continues to change, the role of science in understanding the risks this poses and also the potential changes in exposures faced, cannot be underestimated. By engaging with the scientific community to better understand climate change hazard, exposure, potential loss and risk mitigation, (re)insurers can be ready to help clients navigate this changing landscape, as well as assess, manage and transfer their risks now and in the future.