Advances in catastrophe modeling technology are enabling carriers to take a more microscopic approach to assessing underwriting risks and predicting losses.History does repeat itself. Between 1989 and 1999, insured losses from hurricanes that struck the United States, when adjusted for inflation, totaled $45.7 billion, according to Insurance Services Office Inc. (ISO), New York.
Until recently, carriers did not use-or have much faith in-catastrophe modeling technology. But recent technical advances are helping carriers better manage their corporate risk portfolios to avoid the cataclysmic underwriting losses suffered during the past decade.
Catastrophe modeling advances have also helped the industry develop a way to assess and manage earthquake risks. The technology is gaining wider popularity among carriers that are using the technology to examine risk portfolios, predict losses from natural disasters, deploy claims adjusters in areas afflicted by natural disaster, underwrite property policies and assist in negotiations with reinsurers.
Furthermore, advances in products and technology are enabling carriers to take a more microscopic approach to assessing underwriting risks and predicting losses before a natural disaster strikes. Insurers can look more closely at territories and at individual risks, thanks to the more sophisticated capabilities of "cat" models.
Proving their worth
Carriers have been using catastrophe modeling technology for more than a decade. Typically, early models consisted of databases and computer programs designed to analyze information on potential damages and insurance or reinsurance company exposure from catastrophes such as earthquakes and hurricanes.
Today, however, models use sophisticated statistical analysis and probability factors to estimate losses in areas prone to natural disasters. That information, coupled with other factors, is used by property underwriters to set premium rates.
Cat models proved their worth in 1999 when Hurricane Floyd slammed into the Eastern seaboard, insurers and reinsurers say.
At CNA Re, models helped the company develop a fast and accurate loss estimate. The Chicago-based reinsurer keeps its modeling programs up-to-date by running hypothetical events against its existing portfolio every six months. These tests gauge possible losses in various hurricane-style events-reflecting the company's current risk profile, says John Beckman, vice president, CNA Re.
"When a storm such as Floyd hits, we look at its statistics-its windspeeds, its class and where it will probably hit landfall. Then, we pull out data on all storms with similar characteristics to give us some sort of estimate of what we think the losses will be," he says.
When catastrophe modeling firms release their prognostications about an approaching storm, CNA Re can then refine its own estimates. Within 48 hours of when Floyd began to embark on its destructive trek in the Caribbean Sea, CNA Re knew it probably faced losses of less than $25 million, Beckman says.
"We took a lot of comfort knowing that we were not looking at something that was going to be a big problem in the third quarter," he says. "We knew Floyd was going to cost us some money, but we were able to put a range on it that looked reasonable."
Helping hand
Several major property and casualty insurers put their catastrophe models to work during Hurricane Floyd to make sure claims adjusters were in the right place at the right time.
State Farm Insurance Cos. managed to keep up with Floyd's movement by matching the hurricane's trail against the company's models so that catastrophe teams could be deployed as close to the damaged areas as possible.
Bloomington, Ill.-based State Farm used more than one modeling program-as did most major carriers-to evaluate the storm's progress, says Jeff McCarty, actuary at State Farm.
"We didn't want our people in harm's way, but we wanted them close enough so that they could react as quickly as possible," McCarty says.
State Farm paid more than 105,000 claims totaling in excess of $248 million for Floyd, which caused almost $2 billion in total insured losses, ISO statistics indicate.
Positioning adjusters was just one of the ways that catastrophe models helped insurers respond to Floyd. Allstate Insurance Co. used its models to gain a quick and accurate estimate of how much money it would need to pay property claims, says Dennis Fasking, director of mitigation for the Northbrook, Ill.- based carrier.
Once the storm had passed, adjusters sized up the damage and notified the home office that the models' estimates were on target. "We didn't have to shift much into liquidity," he says. "The models gave us a good idea of how much was necessary and helped us make decisions."
At Travelers Insurance Cos., underwriters can now put hurricane-prone territories under a stronger microscope and pinpoint areas in a state as small as Rhode Island. Catastrophe modeling helps the company to determine if rates are adequate for property coverage in areas that are closest to the coastline and at greater risk of hurricane damage.
A year ago, Travelers would not have been able to look quite that closely, says Ed Charlebois, vice president for personal lines for the Hartford, Conn.-based carrier. Underwriters can now see that rating territories on the coast of Rhode Island have a higher exposure, and they need to analyze those risks more closely than just looking at the state's hurricane risk as a whole, Charlebois says.
Assessing earthquakes
Insurance companies now receive updated hurricane data via the Internet from cat modeling suppliers, enabling carriers to alter their reinsurance portfolios before a storm hits land. However, geologists cannot give carriers advanced warnings about when or where an earthquake will occur.
Nevertheless, cat modeling has become an important tool in helping carriers set policy rates in quake- prone areas.
As the world's largest insurer for personal lines earthquake coverage The California Earthquake Authority (CEA) is perhaps the biggest user of quake-related catastrophe modeling programs. The authority, an insurance pool currently representing about two-thirds of the total earthquake coverage market in the state, was formed in 1996 following the Northridge earthquake that caused about $12.5 billion in losses in 1994.
Computerized modeling is a major part of CEA's procedures for rating quake coverage through simulation of potential losses for events of different magnitude in various areas of the state, says Mark Leonard, a CEA spokesman.
"It's critical for us because unlike other kinds of risks such as hurricanes and tornadoes, earthquakes simply don't have the same kinds of characteristics," he explains. "You can't just sit back and look at actuarial tables."
Pricing property rates
The Northridge quake, which exhibited a series of unusual shaking and damage patterns that had not been seen before, demonstrated that carriers cannot rely on their loss histories as a basis for rate-making, Leonard says.
The CEA's cat modeling programs help the authority to price property rates throughout California. State law requires the authority to use the "best available science" to price coverage and predict losses.
The state is divided into 14 zip-coded territories for which rates are determined by capability of the model to factor in a building's construction type and method, age, the type of soil on which it's situated and its proximity to faults.
EQECAT Inc., an Oakland, Calif.-based provider of risk management software and services, has licensed its USQuake software to the CEA. The program incorporates data from sources such as the U.S. Geological survey and the California Department of Conservation's Division of Mines and Geology in determining quake risk for a given rating area.
New information-such as discovery of new fault zones, revised estimates of anticipated eruption magnitude, and the likelihood of a return quake-is merged into the model once it passes a peer review process. New data is usually added every 12 to 18 months.
These updates have led to an overall decrease of about 15% in the average rate for residential quake coverage over the CEA's three-year history, Leonard says.
In the aftermath of an earthquake, the model helps CEA determine where the greatest ground motions occurred, providing a first glimpse of where to expect claims. "We'll know whether those concentrations are in heavily populated areas or less densely populated areas, which gives us an indication of what our eventual losses will be," he says.
Models weren't used
Although catastrophe models existed when the Northridge quake struck, most of the current capabilities had not been developed. Insurers used cat models infrequently, and those that did use them were still shocked by the loss volume generated by quakes.
Los Angeles-based Farmers Insurance Group, a CEA member, was using a largely manual system for rating quake coverage based heavily on professorial studies of projected losses, says Bob Downer, chief actuary and corporate risk manager at Farmers.
"We weren't alone; a lot of other companies were doing it the same way. There were models available but they just weren't being widely used at all," he explains.
Farmers now uses extensive modeling programs to manage its earthquake and hurricane risks. Catastrophe modeling "is light years ahead of where it was-there's no comparison," Downer says. The company uses the software in California to monitor the rate-making accuracy of the CEA. It also conducts site-specific analysis of commercial quake risks, which the authority does not insure.
Cat models enable Farmers to design product options, such as different deductibles for different coverage levels. Furthermore, cat modeling technology enables the insurer to use its capital more effectively-shifting funds as necessary so they're on hand to pay claims following a catastrophic event.
Internet connection
As is the case with other insurance-industry tools, cat modeling is migrating to the Internet, a development that is providing greater speed and accuracy.
For example, Risk Management Solutions last year launched "RiskOnLine," a service that provides more timely tracking of hurricane movement. The product is intended to enable insurers, brokers and reinsurers to make critical decisions about claims personnel or reinsurance adjustments more quickly.
"RiskOnLine provides reinsurers real-time ability to track hurricane risk," says Hemant Shah, president and co-founder of the Menlo Park, Calif.-based company.
As a hurricane begins rumbling toward land, RiskOnLine provides an analysis of each client's portfolio every three hours as the system is updated with the storm's progress by the National Weather Service.
Such analysis will inform an insurer that it has a 20% chance of sustaining a $100 million loss, for example, if the storm follows its projected path, explains John DeMartini, vice president and coordinator of risk management services for Towers Perrin Reinsurance Services, a New York-based intermediary and consulting company that uses the product.
But if the storm takes a turn and picks up speed a few hours later, the next update may demonstrate a higher percentage of probable losses that could reach $150 million, DeMartini explains. A carrier may then conclude that it needs additional reinsurance and, with damage estimates coming at quicker intervals, can have its broker shop the market and find the additional limits well in advance of the loss.
RiskOnLine and other Internet-based cat modeling products gauge the expected losses with much greater accuracy than was previously possible. In the past, "you had to go back to a theoretical event set of information that contained thousands of events to try and find the one or two that looked the most like the one that was out at sea," DeMartini explains. "You couldn't use the actual meteorological conditions."
John Maes is a freelance writer based in Elk Grove Village, Ill.
