Catastrophe modelers, through their near-term models, have overshot actual insured losses for 2006 through 2010 by as much as $53 billion, Ms. Clark concluded in a recent report. Ms. Clark is president and chief executive of Karen Clark & Co., a risk and risk management firm.

AIR said it has taken an "ensemble approach" to offering views about risk with two models—one based on hurricane activity in all years since 1900 and another derived from years since 1900 with warm sea surface temperatures (WSST). Neither is designed to forecast activity of losses for upcoming seasons, as near-term models attempt to do, AIR said in an e-mail.

Both are probabilistic views of hurricane risk and are each "scientifically sound." Although the WSST model has "inherently more uncertainty" since it is based on half as many years, it has been "peer-reviewed by experts in the field and subsequently published in the Journal of Applied Meteorology and Climatology," AIR said.

Using a probabilistic forecast means that "on average, over many five-year periods, this is the number of hurricanes that can be expected based on the latest research and understanding," RMS said.

"Hurricane losses are extremely volatile on a year-to-year basis, and even over a stretch of several years it is very difficult to draw conclusions about the validity of any model, near-term or long-term," said David Smith, senior vice president of EQECAT's model development group, via e-mail.

Mr. Smith said EQECAT's models are not a prediction of what will happen but a "representation of risk" dependent on aspects in the climate system which are favorable for tropical storms.  

The thought that an increase in storm frequency is due, at least in part, to better observational technology is not a new concept, said RMS, but their models account for this recognition.

Ms. Clark's theory of a "hurricane frequency paradox"—having to do with the relationship between the frequency of hurricanes and landfall—remains open to debate, said RMS, adding that it "does not simply translate Atlantic activity into landfalls when determining its medium-term rates."

"Choosing where to start the clock is critical," said Mr. Smith of EQECAT. "In round numbers, U.S. hurricane losses between 2006 and 2010 averaged about $3 billion per year. However, U.S. hurricane losses between 2004 and 2010 averaged about $15 billion per year."

Per Ms. Clark's report, the three modelers are all providing results around $12 billion to $13 billion per year, Mr. Smith added.