How Did Catastrophe Models Weather Katrina?
While it’s too early to tell if the highly touted computer catastrophe models met expectations for Hurricane Katrina, there are indications the models didn’t work as well as some insurers had hoped.
Several factors-not the least of which was the unexpected flooding and storm surges-have presented challenges to cat models’ accuracy. Some industry observers say that in a best-case scenario, actual insured losses from Katrina will be twice the amount that the models predicted.
But that does not mean the industry will abandon catastrophe modeling.
“If you look at the results, you may think that the models are not doing their job,” said Andrew Castaldi, head of catastrophe and perils for Swiss Re’s America’s Division. “But these are probabilistic scenarios based on speculative events. These will never exactly match the actual event.”
Instead of ditching the technique, providers of cat modeling are taking Katrina as an opportunity to refine or overhaul their programs to account for the powerful storm activity expected to batter the U. S. for years to come.
And while the hurricane showed that total reliance upon cat models is a recipe for disaster, a model integrated into the underwriting process will be a boon for insurers.
Catastrophe models generate different scenarios based on geographic and historical data, and calculate the probability of insured events and potential losses. Carriers and reinsurers use them in two ways: on the front-end, cat models give an idea of what’s at risk, and on the back end, models can help give insurers a quick and accurate picture of exposure following the catastrophe.
Currently, carriers and reinsurers are focused on the back-end aspect of cat modeling, which includes tracking the storm’s impact on aggregated policies and inputting claims data.
Already, insurers are announcing major swings in post-storm estimates. Initial estimates when Hurricane Katrina made landfall two weeks ago were insured losses of $15 billion to $25 billion, according to the various catastrophe modeling firms such as Equecat, AIR Worldwide and Risk Management Services.
But on Sept. 10, RMS revised its loss estimates to as high as $60 billion, making Katrina three times as costly as Hurricane Andrew, which had been the most expensive storm ever in terms of insured losses. Neither Equecat nor AIR have updated their loss estimates.
One major hurdle for the various models was nature and aftereffects of Hurricane Katrina. Much of damage was the result of flooding and a “storm surge” from the Gulf of Mexico, which some models did not take into account. (A storm surge is when coastal water rises above normal levels due to wind. These differ from tsunamis or tidal waves, which occur when water is displaced by geographic events, such as earthquakes.)
“Most models are focused on wind damage which, from what I can tell, was not the most significant major issue in this storm,” Castaldi said. Another aspect of the storm that wasn’t necessarily modeled were losses from business interruption, Castaldi said are likely to be a major part of the overall loss.
And since much of the area that Katrina hit hardest was underwater, cat modeling firms were delayed in their efforts to obtain updated information for their systems, said Jeffrey Berg, senior analyst in the insurance practice of Moody’s Investor Services.
Many weather stations were either flooded in New Orleans or were destroyed by winds in the Gulf of Mexico. These stations help measure information on the exact path of the storm, as well as wind speed. “Many firms had data issues early on since most of the [weather] equipment got flooded or was destroyed,” Berg said.
Jayanta Guin, AIR’s vice president for research and modeling, said initial estimates accounted for the extensive flood damage in New Orleans. While AIR’s model does take into account storm surge, it does not include data on inland flooding, since most insurers do not cover the risk.
“There are a set of issues that are rather complicated once a major city becomes flooded,” Guin said. Although hurricane coverage is usually limited to wind damage, flooding losses may end up being insured or may be contested in court.
Berg said that how efficiently insurers used models to determine the “risk aggregation” to the Gulf Coast, and subsequent purchase of reinsurance to offset that risk, will be the true test of the catastrophe models.
“The question they will ask is how closely the model losses predicted in the front end match up against actual losses,” Berg said.
Swiss Re’s Castaldi said that based on last year’s hurricane season, actual losses to a company’s portfolio were significantly higher or lower than those predicted by the catastrophe models typically because of poor data or non-modeled aspects.
“In many instances, local [insurers] using the models were off by a factor of two or three,” he said. “In the proper hands, cat models can come up with good estimates, but if the data is not correct, it can cause significant problems. There is a lot of human error involved.”
Modeling firms also depend on accurate information, said AIR’s Guin. He explains that while AIR is constantly updating “event sets” to its clientele, it is also using insurers’ claims information to make models more robust.
Pat Tuefel, a principal with KPMG’s Actuarial Services practice, said that catastrophe models improve with each storm. “The insurance industry has known, for quite some time, that there was potential for a significant hurricane through the Gulf to New Orleans and the Gulf Coast,” she said. “The magnitude of the actual event has also been acknowledged as a possibility, through the modeling that has occurred. Any catastrophic event helps to refine the models going forward.”
Guin agrees, and adds that Katrina may change their models even more.
“It will change how modeling is done,” he said. “Over the next year we will collect a lot of claims data, weather data and scientific analysis. It will be interesting to see.”
Christopher Westfall is managing editor of
KPMG’s Insurance Insider. Article reprinted
with permission from KPMG’s Insurance
Insider. Copyright 2005 KPMG LLP. All rights
reserved. Disclaimer from KPMG: All informa
tion provided is of a general nature and is not
intended to address the circumstances of any
particular individual or entity.
Instead of ditching the technique, providers of cat modeling are taking Katrina as an opportunity to refine or overhaul their programs.