Catastrophe modelers are working on building climate change effects into their models amid criticism from top insurance executives that their products fail to account for the impact of global warming.
But they also warn that, although climate change is irrefutable, it is difficult to measure, and tough to definitively blame for the recent rise in natural catastrophe activity.
Insurance executives' awareness of climate change has been raised by three years of heavy catastrophe loss activity, which has had profound effects on the competitive and financial dynamics of their industry. The 2019 catastrophes — chiefly Hurricane Dorian and typhoons Faxai and Hagibis — followed a combined natural catastrophe bill of $219 billion for 2017 and 2018, according to Swiss Re.
Reinsurance prices have started to increase in response to the catastrophe claims bill, but although higher prices present the opportunity for profitable growth, some executives have clearly been troubled by what they are seeing. Kevin O'Donnell, chief executive of Bermuda-based reinsurer RenaissanceRe Holdings Ltd., told analysts on an Oct. 30 earnings call that he was not convinced that models yet reflected the "new reality" of climate change's effect on natural catastrophes.
Reinsurers typically use such models to determine the likelihood of natural catastrophes in the coming year, helping insurers decide what cover to buy and reinsurers what price to charge. Because traditional statistical catastrophe models rely heavily on historical data, any climate change impacts to date are inherently built in, but pressure is now mounting to take a longer-term view, and consider what might happen in the next several decades as well as just the following year.
The U.K.'s Prudential Regulation Authority, for example, now expects insurers to assess, manage and report on the financial risks of climate change, and more regulators are expected to follow suit.
Adam Podlaha, head of Aon PLC's Impact Forecasting catastrophe modeling division, said traditional insurance usage did not require models to consider what the weather would be like in, for example, 50 years' time. But thanks to regulators' focus on sustainability and resiliency, "the cat models are also being asked to do this sort of job."
The missing link
The problem is that the link between climate change and more frequent and severe catastrophes is far from clear. Climate change may be visible on a global scale, but catastrophe modelers require evidence at the country, regional or even more local level, according to Arno Hilberts, vice president of model development at risk modeler RMS.
"That level of detail is often not available or it is completely unclear what it would look like at that level," he said.
The frequency and severity of Atlantic basin hurricanes are heavily influenced by recurring climatic phenomena that affect sea-surface temperatures, such as the El Niño-Southern Oscillation and the Atlantic Multi-decadal Oscillation, masking any climate change influence.
Peter Sousounis, vice president of climate change research at risk modeler AIR Worldwide, noted that, amid the noise of other phenomena, the signal from climate change was relatively weak because the phenomenon has occurred gradually since the middle of the 19th century. The impact from El Niño "is much greater than any signal that climate change is providing, if it is providing any at all right now," he added.
Historical trends, meanwhile, can be distorted by changes such as improvements in wind-speed measuring technology, Sousounis said, adding that using older data, when the climate might have been different, could mean "diluting the current climate signal." Equally, he said, completely excluding older years may throw out valid information on event variability.
The absence of hard evidence means that some modelers are taking a conservative approach to updating their models. Hilberts said RMS was wary of presenting clients with an ever-shifting baseline view of the risk in response to new research on climate change.
"We only factor in climate change impacts if we are absolutely certain they are there and also that we know how to factor them in," he said.
Yet neither are risk modelers ignoring climate change. Karen Clark, CEO of catastrophe modeling firm Karen Clark & Co., said scientists believe that climate change could be causing more "wet" storms — those that bring heavy rains and cause more flood damage, such as 2017's Hurricane Harvey and 2019's Typhoon Hagibis.
She said her company was working on adding a wet storms component to its hurricane model, and it should be included in 2020.
For severe convective storms — typhoons, thunderstorms and hailstorms — Clark's firm is shifting from statistical modeling to physical modeling of the atmosphere, enabling the company to focus on the most recent 10 years of data.
"What that means is that any changes in the atmosphere caused by climate change ... automatically become part of our model," she said.
Physical modeling work is also underway for wildfires and winter storms, although Clark added that physical modeling was not currently applicable to hurricanes, which are smaller, and where greater resolution is required.
AIR Worldwide is limiting the historical data it uses to the past 40 years for many of its models, Sousounis said. The company also plans to make some model changes in response to certain trends, such as the eastwards shift in "Tornado Alley" in the U.S.
And Hilberts said RMS is considering whether to produce views on what a future climate could look like for its clients to experiment with.
Ultimately, the insurance industry may need to adapt to looking at several possible versions of the future when using models to understand climate change impacts. Clark said: "The industry will have to get used to looking at different potential scenarios."