How insurance companies calculate catastrophe

A look at the methods the industry uses to adapt to climate change, which portends more storms and greater risk

Published September 29, 2013 6:00PM (EDT)

This article originally appeared on

Smithsonian Magazine When it comes to the calculating the likelihood of catastrophic weather, one group has an obvious and immediate financial stake in the game: the insurance industry. And in recent years, the industry researchers who attempt to determine the annual odds of catastrophic weather-related disasters—including floods and wind storms—say they’re seeing something new.

“Our business depends on us being neutral. We simply try to make the best possible assessment of risk today, with no vested interest,” says Robert Muir-Wood, the chief scientist of Risk Management Solutions (RMS), a company that creates software models to allow insurance companies to calculate risk. “In the past, when making these assessments, we looked to history. But in fact, we’ve now realized that that’s no longer a safe assumption—we can see, with certain phenomena in certain parts of the world, that the activity today is not simply the average of history.”

This pronounced shift can be seen in extreme rainfall events, heat waves and wind storms. The underlying reason, he says, is climate change, driven by rising greenhouse gas emissions. Muir-Wood’s company is responsible for figuring out just how much more risk the world’s insurance companies face as a result of climate change when homeowners buy policies to protect their property.

Climate change could mean more frequent wind storms, increasing the level of risk taken on by insurance firms. Photo by Flickr user PSNH

First, a brief primer on the concept of insurance: Essentially, it’s a tool for spreading risk—say, the chance your house will be washed away by a hurricane—among a larger group of people, so that the cost of rebuilding the destroyed house is shared by everyone who pays insurance. To accomplish this, insurance companies sell flood policies to thousands of homeowners and collect enough in payments from all of them so that they have enough to pay for the inevitable disaster, plus keep some extra revenue as profit afterward. To protect themselves, these insurance companies even buy their own policies from reinsurance companies, who make the same sorts of calculations, just on another level upward.

The tricky part, though, is determining just how much these companies need to charge to make sure they have enough to pay for disasters and to stay in business—and that’s where Muir-Wood’s work comes in. “If you think about it, it’s actually quite a difficult problem,” he says. “You’ve got to think about all the bad things that can happen, and then figure out how likely all those bad things are, and then work out ‘How much do I need to set aside per year to pay for all the catastrophic losses that can happen?’”

With natural disasters like floods, he notes, you can have many years in a row with no damage in one particular area, then have tens of thousands of houses destroyed at once. The fact that the frequency of some catastrophic weather events may be changing due to climate change makes the problem even more complex.

The best strategy for solving it is the use of computer models, which simulate thousands of the most extreme weather disasters—say, a record-setting hurricane slamming into the East Coast just when the power grid is overloaded due to a heat wave—to tell insurance companies the worst-case scenario, so they know just how much risk they’re taking on, and how likely it is they’ll have to pay out.

“Catastrophes are complex, and the kinds of things that happen during them are complex, so we are constantly trying to improve our modeling to capture the full range of extreme events,” Muir-Wood says, noting that RMS employs more than 100 scientists and mathematicians towards this goal. “When Hurricane Sandy happened, for instance, we already had events like Sandy in our models—we had anticipated the complexity of having a really big storm driving an enormous storm surge, even with wind speeds that were relatively modest.”

These models are not unlike those used by scientists to estimate the long-term changes our climate will undergo as it warms over the next century, but there’s one important difference: Insurance companies care mainly about the next year, not the next 100 years, because they mostly sell policies one year at a time.

But even in the short term, Muir-Wood’s team has determined, the risk of a variety of disasters seems to have already shifted. “The first model in which we changed our perspective is on U.S. Atlantic hurricanes. Basically, after the 2004 and 2005 seasons, we determined that it was unsafe to simply assume that historical averages still applied,” he says. “We’ve since seen that today’s activity has changed in other particular areas as well—with extreme rainfall events, such as the recent flooding in Boulder, Colorado, and with heat waves in certain parts of the world.”

RMS isn’t alone. In June, the Geneva Association, an insurance industry research group, released a report (PDF) outlining evidence of climate change and describing the new challenges insurance companies will face as it progresses. “In the non-stationary environment caused by ocean warming, traditional approaches, which are solely based on analyzing historical data, increasingly fail to estimate today’s hazard probabilities,” it stated. “A paradigm shift from historic to predictive risk assessment methods is necessary.”

Moving forward, Muir-Wood’s group will attempt to keep gauging the shifting likelihood of a range of extreme weather events, so that insurers can figure out how much to charge so that they can compete with others, but not be wiped out when disaster strikes. In particular, they’ll be closely looking at changing the model for flooding rates in higher latitudes, such as Canada and Russia—where climate is shifting more quickly—as well as wildfires around the planet.

On the whole, it seems likely that insurance premiums for houses and buildings in flood-prone coastal regions will go up to account for the shifts Muir-Wood is seeing. On the other hand, because of the complex impacts of climate change, we might see risks—and premiums—go down in other areas. There’s evidence, for example, that snowmelt-driven springtime floods in Britain will become less frequent in the future.

For his own part, Muir-Wood puts his money where his mouth is. “I personally wouldn’t invest in beachfront property anymore,” he says, noting the steady increase in sea level we’re expecting to see worldwide in the coming century, on top of more extreme storms. “And if you’re thinking about it, I’d calculate quite carefully how far back you’d have to be in the event of a hurricane.”


By Joseph Stromberg

MORE FROM Joseph Stromberg

Related Topics ------------------------------------------

Climate Change Hurricanes Insurance Insurance Companies Risk Storms