“They [modelers] are looking at exactly the same data, but very little is available,” Clark says. “The modeling is highly sensitive to any change in assumptions.”

Overestimate wind speed by a couple of miles-per-hour and the loss estimate can change 50-to-100 percent, offers Clark as an example.

In other words, modelers are applying different assumptions off of few data points. Therefore, estimates of insured losses following an event often differ, but in the case of Irene, the difference is a bit eye-opening.

“We're a little confused as well,” says Tom Larsen, senior vice president of Eqecat, when asked why AIR's insured-loss estimate was so much more. “We can only speculate.”

Eqecat looks at maximum wind speeds, the radius of the wind and how fast the storm moves. It also uses its insured exposure database (IED) to determine home values and insurance conditions—homes with federal flood insurance and wind deductibles—in affected areas, Larsen says.

The process is fairly complex, he says, while he admits that slight overestimations could have a substantial effect on damage results.

AIR spokesman Kevin Long says the stark difference in insured loss estimates “further reinforces the point that not all models are the same.” He too referenced the data.

“There can be differences in the approach/methods to modeling and how certain building types perform under different wind speeds,” Long says. “Exposure data is vital—the quality and completeness of the data used.”

Clark says her company has come up with a $3 billion loss estimate for the insurance industry after using RiskInsight, a new tool the company developed.