Shaky Loss Estimates Often Follow Catastrophes

Catastrophe modeling firms will continue to issue quick loss estimates following major catastrophes even though they often significantly underestimate the ultimate costs of a disaster.

While the estimates are often off by factors of two or more, insurance industry clients are pushing modelers to provide immediate estimates so they can assuage shareholders’ and competitors’ concerns about their ability to weather a particular event.

There is a joke among catastrophe modeling firms that some CEOs want to know the losses before a hurricane makes landfall,” says Karen Clark, president of Boston-based consulting firm Karen Clark & Company.

Shortly after two major disasters last week — the earthquake in Chile and European wind storm Xynthia — some of the largest catastrophe modeling firms quickly came out with loss estimates.

On the same day a magnitude 8.8 earthquake struck Chile, Eqecat estimated total economic losses from the event would range between $15 billion to $30 billion. Several days later, the firm estimated that insured losses would range between $3 billion to $8 billion, or 25 percent of economic losses.

Boston-based AIR Worldwide issued a press release a day following the disaster estimating that economic losses from the quake could exceed $15 billion, with insured losses topping $2 billion.

Following the destruction caused by wind storm Xynthia in France, Belgium, Germany and the Netherlands, AIR estimated insured losses would range between 1.5 billion and 3 billion Euros.

Catastrophe modeling firms issue initial estimates to quench the thirst of insurance industry executives and stock analysts looking to understand the impact of catastrophes on their business and finances, Clark says.

Often, executives want a heads-up on losses from a particular storm or earthquake to get a sense of how the event can affect shareholder value and the firm’s bottom line.

But initial estimates are often far off the mark from actual losses, and industry professionals need to include that uncertainty into their planning, Clark argues.

“The first estimates of losses during the 1994 Northridge Earthquake were $2 billion, and 18 months later they were up to $12 billion,” she says. “In Chile it could take even longer [for more accurate numbers] because there is far less information on insured values.”

Because of the uncertainty around initial loss estimates, firms such as RMS often decline to issue statements too close to the time of an event.

“It’s RMS policy to refrain from issuing industry loss estimates while the impacts of an event are still unfolding,” the firm said in an email statement.

However, insurers are sophisticated users that understand the uncertainty baked into the estimates, modelers argue.

A lot of uncertainty that goes into risk modeling, such as how the ground shakes, how structures will respond and how much that damage will cost,” says Katherine Stillwell, manager of earthquake product for modeling firm Eqecat. “There is a great deal of uncertainty in all the components.”

Stillwell says Eqecat follows its own Chile earthquake methodology comprising 15 different total asset values that are then imported into the Chilean model, which separates the country into distinct “zones.”

She adds the Eqecat measurement of Chilean earthquake losses are based on the actuarial mean of potential losses plus a standard deviation of one.

Clark agrees that the most important issue when it comes to initial loss estimates is the vast array of information that is not included

“The reason there are so many different loss estimates is that there are a huge number of unknowns,” says Clark.

For example, Clark explains little is understood about the actual ground motion throughout the region of Chile where the quake occurred, and there is even less information regarding the value of insured property at each affected location.

Additionally, catastrophe modeling firms do not have information on the exact wind speeds at every location in Europe that experienced last week’s windstorm.

“You just can’t expect catastrophe models to know these things,” Clark says.

Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to Risk Market News.

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.