Cat model updates increase demand surge assumptions

The catastrophe models in 2007 have updated their demand surge assumptions and industry exposure levels, resulting in an increase in estimated average annual losses, according to Fitch Ratings Hurricane Season 2007: A Desk Reference for Insurance Investors.Because catastrophe losses fell dramatically in 2006 from the previous two record-setting years, changes to the major hurricane models have been more incremental in contrast to the significant changes made to the models in 2006, the report said.Risk Management Solutions (RMS) contended that the use of long-term averages underestimates the level of hurricane risk and is inconsistent with the opinions of expert hurricane researchers, according to the report. In its 2007 model, RMS increased its demand surge assumptions and industry exposure levels with a net result of a 14% increase in the estimated annual average loss and a 22% increase in the 1-in-100-year loss.Unlike RMS, AIR Worldwide Corporation (AIR) continues to base its modelled view of storm activity from a longer-term perspective, stating that historical data represents an essential element in the projection of hurricane activity.As well as updating its standard stochastic catalogue, based on more than 100 years of historical data, AIR continues to offer an alternative stochastic catalogue for 2007 that contemplates the influence of ocean-warming trends on medium-term (approximately 5 years) hurricane activity, the researchers at Fitch reported. Updates in vulnerability functions and changes in the way demand surge is applied resulted in a net effect of a 5% – 7% increase in estimated average losses.EQUECAT International Inc. (EQE) also offers two views of hurricane risk, one based on the long-term and the other the near-term.EQEs changes to its 2007 model include: updating the model to include 2006 activity; updating land use data outside Florida; and updating vulnerability and demand surge outside Florida.According to Fitch, EQE expects these changes to produce very small differences from the model used in 2006.