During the call, Mr. Fischer, who is the primary insurance marketplace leader for the Boston-based catastrophe modeling company, said the widespread notion that the models did not work last year was not quite accurate.

"AIR's review of the data and benchmarking of their estimates against underwriters' reported losses indicates that their models were reasonably accurate," Mr. Fischer said on the call.

Many of the issues that arose were due to incomplete or inaccurate exposure data such as missing locations, incorrect replacement value and incorrect property coding used by underwriters in the modeling process.

"Since the storms, we have heard that underwriters understand the 'garbage in, garbage out' problem and are working to improve the data input into the models," Mr. Small wrote. "A concern for investors is that it is difficult to independently assess the progress made."

Even without the increased risks associated with higher hurricane frequency and severity, aggregate loss values are rising dramatically due to factors such as heavy development in the most catastrophe-exposed areas, Mr. Small wrote.

"Looking forward, continued exposure growth alone could cause a doubling of loss expectancies from a given hurricane over the next decade," he wrote.