RMS Reacts to Charges of Faulty Cat Models

Newark, Calif. - Risk Management Solutions (RMS) a provider of products and services for the quantification and management of catastrophe risks, scrambled to respond to charges leveled by the Tampa Tribune that its CAT models rely on "faulty science," and that insurance companies are using the models to justify huge rate increases in coastal areas. In a letter to Insurance Networking News, Dr. Robert Muir-Wood, chief research officer of RMS, Newark, Calif., disputed the January 7 article entitled "Insurance Risk Forecast Called Faulty," stating, "We at RMS were stunned by the article's inaccuracies and one-sidedness." Contributing to the dispute is a change made in March 2006 to the RMS model that takes a 'medium-term' (five-year) forward-looking view of risk for estimating potential catastrophe losses. To date, catastrophe model results have typically been based on a long-term historical average baseline. Jim Elsner, a professor of geography at Florida State University and one of four experts on a panel that provided input to the model's development, criticized the results, telling the Tampa Tribune that it contains assumptions that are "actually unscientific." In response, RMS confirmed convening two separate meetings one with Elsner and one in which Elsner, citing an affiliation with an RMS competitor, was absent. In the October 2005 meeting, RMS hosted a meeting of four hurricane climatologists to develop a consensus forecast of the overall level of U.S. hurricane activity expected over the next five years.  "The consensus involved weighing the opinions of the four experts, having first provided them with detailed statistics on historical hurricane activity and landfalls. All four scientists, including Professor Elsner, gave their sign-off on the outcome of this process. RMS then took the results of this forecast and implemented them in its hurricane catastrophe model," Muir-Wood told INN. "RMS climatologists took responsibility for determining where the extra hurricanes would be expected to form, while preserving the overall target activity rates established by the expert panel," he said. "A press release and white paper were issued describing this work in detail. Again, all four experts were asked to review and approve both documents, to ensure that their involvement was appropriately represented. The RMS regional landfall rates were not challenged by any of the panelists. While it is now recognized that Professor Elsner has developed his own theories on how hurricane activity translates to regional landfall rates, he did not challenge the RMS landfall rates developed after the 2005 expert elicitation." As reported on March 23, 2006 in Insurance Networking News, RMS justified its updated five-year model, which predicted an increase in modeled annualized insurance losses by 40% on average across the Gulf Coast, Florida and the Southeast. Modeled annualized insurance losses in the Mid-Atlantic and Northeast coastal regions will increase by 25% to 30%, relative to those derived using long-term 1900-2005 historical average hurricane frequencies, INN reported. "This new view of risk is driven by an increase of more than 30% in the modele frequency of major (Saffir-Simpson Category 3-5) hurricanes making landfall in the U.S. to account for current elevated levels of hurricane activity in the Atlantic basin, which are expected to persist for at least the next five years. When compared with a pre-2004 historical baseline, as has been previously employed for quantifying insurance risk, the increases in modeled annualized losses are closer to 50% in the Gulf, Florida, and the Southeast," reported INN. Taking into account that eight storms have hit the area in the last two years, RMS stated that the increased frequency and intensity of hurricane activity in the Atlantic Ocean Basin, as observed since 1995, is driven by higher sea surface temperatures in the tropical North Atlantic and by associated changes in atmospheric circulation. These warmer temperatures are expected to translate into a continuation of high activity in the basin, leading to a greater potential for hurricanes to make landfall at higher intensities over the next five years. "Hurricane activity in the Atlantic basin has been running far above the historical average in 9 of 12 years since 1995," Muir-Wood told INN. "As one metric, the annual number of the most intense storms (Category 3-5) has been more than twice that of the average annual number between 1970 and 1995." In October 2006 RMS organized the second of these annual expert meetings on hurricane activity rates, inviting all four of the scientists involved in the first meeting, reports RMS.  "Only Professor Elsner declined, citing that he was under contract with a company affiliated with our main competitor," Muir-Wood told INN. The second meeting involved a total of seven climatologists, and went into greater depth than the first meeting, employing the results of 20 different statistical and climatological forecasting methods, said Muir-Wood.  "The activity rate forecast for the next five years that came out of this meeting was almost identical - within 1-2% - of the projection of the first year's meeting," Muir-Wood stated. Despite the charges leveled by Elsner, RMS will continue to run an annual hurricane climatology expert elicitation procedure to ensure that RMS hurricane models reflect the most current view of hurricane risk, said Muir-Wood.  "The five year perspective, may in future years, be decreased if this is suggested by the best scientific and statistical evidence available at that time," he added. Since the January 7 Tampa Tribune article appeared, further coverage in the popular press has linked questions about the catastrophe models to the rate increases being employed by insurance companies. Yesterday, two Florida Cabinet officers asked for more information about "a dramatic change in hurricane damage forecasting used by the insurance industry." In the Tampa Tribune's January 9 edition, Gov. Charlie Crist and Chief Financial Officer Alex Sink challenged RMS to provide the background for their model creation.  "All of the material produced at these [annual expert] meetings, as well as the details of how activity rates were implemented, have been documented and are in the process of being published in peer-reviewed scientific literature," Muir-Wood told INN. RMS is the official model for the Florida Hurricane Catastrophe Fund, which was created after Hurricane Andrew hit in 1993. The fund provides backup coverage for private insurance companies.  "RMS has built its reputation on the principles of providing neutral and unbiased information on risk," concluded Muir-Wood. Sources: Tampa Tribune, The Kansas City Star, Risk Management Solutions, Insurance Networking News archives.