Insurance & Technology is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Policy Administration

12:49 PM
Connect Directly

One Year After Katrina: Improved Technology Helps Insurers Weather the 2006 Hurricane Season

Insurers assess risk of this year's Atlantic hurricane season with improved catastrophe modeling and risk decision geographic information systems technologies as well as better property and catastrophe data management.

One year after Hurricane Katrina struck Louisiana and Mississippi, the U.S. insurance industry will find out if the technology they have invested in will stand them in good stead for this year's Atlantic hurricane season, beginning with Tropical Storm Ernesto. The National Oceanic and Atmospheric Administration (NOAA; Washington, D.C.) is projecting a total of 12 to 15 storms, seven to nine of which will intensify into hurricanes for the entire hurricane season, which ends November 30. To prepare for this year's stormy season, insurers have been investing in catastrophe (CAT) modeling software and geo-information systems to mitigate potential catastrophe losses.

"Insurers are taking the technology they had and pushing it to a higher form of usage with underwriting and modeling software that will help insurers deal much more effectively with risk decision making," says Karen Pauli, senior analyst for TowerGroup's (Needham, Mass.) insurance practice.

Citizen's Property Insurance Corp. (Tallahassee, Fla.; $4 billion in premium) has grown from 627,000 to 1.2 million policies in the past year, due to carriers such as Allstate (Northbrook, Ill.; $94 billion in assets) and State Farm (Bloomington, Ill.; $56 billion in revenue) pulling out of the Florida market. Citizen's Property Insurance Corp. uses both internal and external catastrophe modeling technology by RMS (Newark, Calif.) to prepare for storms such as Ernesto.

"We've learned from experience, and we have been watching Ernesto," says Curt Overpeck, senior vice president and CIO of Citizen's Property Insurance Corporation. "We have been in catastrophe activation mode, which includes our modeling and field deployment plans for three days. We are using CAT modeling to predict losses from this tropical storm and conduct exposure analysis."

Property and casualty insurers -- who suffered $40.6 billion on 1.7 million claims in six states from the damage of Hurricane Katrina alone, according to the Insurance Information Institute (III; New York) -- have learned that higher quality data is the key to protecting itself against catastrophic situations.

"The industry has realized that the data that goes into decision making and product development needs to improve significantly," says TowerGroup's Pauli. "The industry data comes out of legacy systems, and much of it is disparate. Carriers need better access to data in order to pinpoint specific risk and measure exposure."

Andy Castaldi, senior vice president of catastrophe perils at Zurich, Switzerland-based reinsurer Swiss Re (net premium $22.6 billion) concurs that data management was one of the key lessons of Hurricane Katrina. "Insurers are starting to understand that you need the best information as possible," says Castaldi. "A lot of what we believed to be proper assumptions were blown away with a substantial amount of property and losses. We though properties were less vulnerable and damage would be less."

Many of these assumptions were based on vendor models built from 100 to 150 years of storm data that may no longer be accurate for use in current models, according to Castaldi. "These models were built when the climate was at a certain level. The temperature of the earth has increased since then, and the last seven storms in the past two years have given us a chance to reevaluate models," says Castaldi.

Catastrophe modeling vendors, such as RMS, AIR (Boston, Mass.), EQECAT (Oakland, Calif.), and Lloyd's of London have been developing models more realistic scenarios. "There has been renewed interest in CAT modeling and a healthy critique in-house to use CAT models," says Josh Darr, senior model manager for North America climate hazard or RMS. "Insurers are bringing the models in house to have a lot more control over the data. When insurers run the models themselves they can get much more accurate loss results."

Many carriers are also using CAT models along with risk-decision models and geo-information models, which pinpoint specific risk locations and are made by vendors such as Group 1 Software (Lanham, M.D.), or with mapping products, such as MapInfo (Troy, New York), to assess risk based on location more effectively, according to TowerGroup's Pauli. "Insurers are using catastrophe modeling and risk modeling in a much more sophisticated manner to heighten [identification of] risks relative to catastrophe zones," she says.

In the past year, St. Paul Travelers (St. Paul, Minn.; 2005 total assets of $75 billion), which uses both RMS an AIR hurricane models, has reevaluated the assumptions it is making on historic data. "We had been using models based upon 100 years of storm data and worked with our vendors to limit that data down to a five year time frame that would reflect warmer ocean conditions to adjust our pricing and evaluate our exposures," says Bruce Jones, senior vice president and chief actuary at Travelers.

Travelers is using CAT modeling to calculate its average annual loss and reflect its exposure, but the carrier is also taking other data into consideration. "I think that Katrina gave us the realization of how fragile models are. Models aren't always measuring certain classes of business and building types well, and insurers really need to step back to review our data quality and measure projected losses relative to actual losses," says Jones.

Further inaccuracies in models were also based on carriers' inability to keep property values up-to-date, misleading interpretation of models, and the absence of secondary risk hazards in models, according to Castaldi. "Most importantly, a number of us have learned that these models are not storm models -- they are wind models," he says. "There are a number of secondary hazards that can contribute to overall storm loss; plus, the information we have been feeding these models has not had up-to-date loss values, locations, or risk distribution, and still we have relied heavily on these models."

On The Net


Register for Insurance & Technology Newsletters