Wall Street Journal: Insurers Criticized For New Rate Models
Jul 1, 2008
Wall Street Journal--July 1, 2008
By M.P. MCQUEEN
Scientists say the jury is still out on whether rising sea temperatures will cause more hurricanes to hit U.S. coastlines. Yet some insurance companies are boosting premiums based on assumptions that they will. Others are withdrawing from coastal communities altogether.
Last year, Leanne Lord of Marion, Mass., decided to put her house up for sale after her insurance premiums more than doubled to about $2,892 a year since 2005. Many of her Cape Cod neighbors, who hadn’t seen a hurricane in the area since 1991, followed suit. Today, there’s a glut of houses on the local market.
“A lot of people can’t afford to live here anymore, between the insurance and the taxes having gone up so much,” says Ms. Lord, a 52-year-old public-health nurse. “They have been forced to leave and I think that is really sad.”
Costs for homeowner insurance along the East and Gulf coasts have risen 20% to 100% since 2004, says the Insurance Information Institute, a trade group. In the three years through 2006, says the institute, property and casualty insurers registered record profits, topping out at $65.8 billion in 2006. (Despite severe U.S. weather that has caused about $8.9 billion in insured property losses to date this year, it’s too early to forecast 2008 profits.)
Helping to drive these developments is a little-known tool of the insurance world: Computerized catastrophe modeling. Crafted by several independent firms and used by most insurers, so-called cat models rely on complex data to estimate probable losses from hurricanes.
But regulators and other critics contend that the latest cat models — which include assumptions about various climate changes — are triggering higher insurance rates.
Starting in the early 1990s, cat models began to replace the industry’s older tools. Previously, insurers based their rates and underwriting policies largely on historical records of past claims. The turning point in methodology came after 1992, when Hurricane Andrew wrought damages in excess of $15.5 billion and left about a dozen insurers insolvent.
The original purpose of cat models was to help stabilize the insurance market and ensure affordable coverage in risky areas. To do this, the first versions used historical weather data to project long-term future losses.
In the wake of the punishing 2004 and 2005 hurricane seasons, many cat models saw drastic revisions. Rather than take a traditional longterm view, some attempted to estimate what might happen in the next several years. Modelers also factored in dramatically higher rebuilding costs when a large area is hit. The result: big premium hikes and higher deductibles.
Underlying the newer cat models are scientific theories that rising sea temperatures will result in more intense, and possibly more frequent, hurricanes. The hypotheses suggest that catastrophic hurricanes like 2005’s Rita, Wilma and Katrina weren’t an aberration, but rather the shape of things to come.
Large reinsurance companies, such as Swiss Re and Munich Re, were early converts to theories of global warming and cite warming of the earth’s oceans when predicting massive damages from future storms.
“Losses from hurricanes and tropical storms have risen along with sea temperatures,” says Eberhard Faust, a climate scientist at Munich Re. “This is [the assumption] from where all the modelers start.”
That sea-surface temperatures are rising is no longer much in dispute. There is also near-consensus that rising temperatures are linked to greater hurricane activity. However, scientists remain divided over how that may affect the number and intensity of hurricanes making landfall in the coastal U.S. A few climate experts believe global warming might actually cause fewer hurricanes to come ashore on the East Coast.
The National Oceanic and Atmospheric Administration has predicted a “normal or above-normal” hurricane season in the North Atlantic, which began June 1. No hurricanes have hit the U.S. thus far. But the agency noted that it could not predict how many storms would make landfall.
Some scientists, like Chuck Watson, contend that cat models may fail to take into account such scientific uncertainty. A geophysicist, he advises the Florida Commission on Hurricane Loss Projection Methodology — a state panel that reviews insurer models. Cat models, he says, use assumptions about variables like air pressure and wind velocity that can’t be known precisely. As a result, the models can be overreaching in their conclusions and “are not being properly applied” by some insurers, he says.
Perhaps the most prominent critic to surface is Karen Clark, an economist who founded one of the first cat-modeling firms two decades ago. Today, she warns about the programs’ misapplication.
After Katrina, she attended insurance-company meetings to discuss “what went wrong” and concluded that there were more problems with how insurers were using the models than with the models themselves.
Companies that rely too heavily on cat-model data “are subjecting their businesses and their customers to the volatility of computer models,” says Ms. Clark, who now runs a Boston cat-model consulting business. “The models are being used as if they produce definitive answers rather than uncertain estimates.” Ms. Clark says she advises clients to use them in conjunction with other factors, such as broad historical data.
To be sure, insurers themselves are facing higher rates from the reinsurance companies that backstop their claims. The reinsurers, and the financial ratings agencies that assess the health of carriers, are also using the controversial newer models.
Insurance companies say their cat models are generally reliable and credit them with sharply reducing the number of carriers that have failed following major disasters in recent years.
The big independent modeling firms are Applied Insurance Research Worldwide Corp. of Boston; Risk Management Solutions Inc., a unit of Britain’s Daily Mail & General Trust PLC; and EQECAT Inc., a subsidiary of the ABS Group of Cos. in Houston. Each sells cat models to insurance and financial-services companies around the world, as well as to governments and other groups. Many insurers also have their own in-house modeling teams.
To construct a hurricane model, programmers factor in a century or more of meteorological data, along with information from insurers such as the replacement value of buildings. The programs can simulate thousands of hypothetical hurricanes and estimate the maximum property losses for each depending on its path.
The modeling concerns say they do their best to make sure their programs are based on sound science and methodology. They also emphasize that in order for their cat models to work, insurers must provide accurate information on the properties they cover.
“They are complicated and rely on the quality of the information that you put in,” says Tom Larsen, senior vice president of EQECAT. Incomplete or inaccurate data about property or weather, he notes, can affect results.
State insurance regulators and consumer groups are beginning to push back, saying that some insurers are relying too heavily on their use of cat models. Such critics note that the industry managed to realize huge profits in recent years — despite record damages from back-to-back hurricanes in 2004 and 2005. (The Insurance Information Institute, the industry’s trade group, says profitable years help offset bad spells when claims can exceed premiums. That happened in 2001, when insurers lost $7 billion.)
In May, Massachusetts officials denied a 25% rate increase that had been sought by the state-administered FAIR plan, its insurer of last resort. The request was “based in large part on a hurricane model that is not calibrated for Massachusetts weather patterns,” state Attorney General Martha Coakley said in a statement. It “predicts the type of storms that have never made landfall in Massachusetts.”
Jack Golembeski, president of the FAIR plan, said it is reviewing the state’s decision to deny a rate increase and would resubmit a new request at another date.
The newer models have caused other skirmishes. Industry officials note that some models now attempt to estimate future losses over a shorter period of time. In doing so, they may also use selective historical data. One model, for example, was reprogrammed to give greater weight to years in which ocean temperatures were particularly warm and hurricane rates were high, such as the period from 1930 to 1945. That particular model resulted in higher loss estimates for the near-term.
J. Robert Hunter, insurance director for the Consumer Federation of America, says that the organization generally supports cat models, but only when they’re used in scientific ways. The longer-term models, he says, ensured “stable pricing so that no hurricanes would not lower prices much and multiple hurricanes would not raise prices much.” After Katrina, he contends, some modelers and their clients “reneged” on promises to stick with the long-term models.
EQECAT’s Mr. Larsen says he is unaware of any pact to use specific models.
Claire Souch, a senior director at Risk Management Solutions, says its model used to be based on the long-term average. “But that doesn’t reflect the risk today because we are in a phase of above-average [hurricane] activity,” says Ms. Souch. RMS introduced the industry’s first five-year model in 2005.
Last fall, Florida rejected Allstate’s request for a 43% rate increase. In the process of investigating the proposed rate hike, officials learned that Allstate’s cat model didn’t comply with the state’s rules against using a short-term model.
“They were using five-year projections [in their cat modeling] even though they had not been approved, and our concern was that it was because it generates a higher loss cost,” said Kevin McCarty, the state’s insurance commissioner.
Allstate acknowledged that it had used a short-term version, but only in conjunction with an approved long-term model. A company spokesman says the insurer didn’t violate Florida laws at the time and that its model provided “a more accurate prediction of risk.” (The state has subsequently passed a law specifically forbidding use of unapproved models.)
The impact from cat models on homeowners along the East and Gulf coasts has stirred some of the greatest controversy. In New Jersey, State Farm Mutual Insurance Co. and a subsidiary of Allstate Corp. have declined to renew at least 12,000 customers with homes near the ocean. In Mississippi, several insurers, including Nationwide Mutual Insurance Co., have stopped covering wind damage in six counties along the Gulf. Some homeowners in the region got a 90% premium increase in 2006. And in Florida, State Farm, the largest private insurer there, said recently it would no longer write new homeowner policies and planned to drop 50,000 existing ones.
Catastrophe modeling has also influenced other types of coverage. In 2006, Allstate stopped writing new earthquake policies across much of the country. One reason for the change: Its cat models estimated billions of dollars in damages if a major quake struck along the New Madrid fault, which runs through the Missouri town of the same name.
All the complexity has prompted some state regulators to employ independent cat modelers and climate experts to assess for themselves the damage projections made by insurers.
Florida has maintained its own cat model since 2005. The state-run program takes a long-term view of hurricane activity in projecting future losses. When large discrepancies arise between the public model and those submitted by insurers, it triggers further inquiry. South Carolina has employed the services of the Florida public model to analyze its own insurers’ rate requests.
Meanwhile, consumers are often caught in the middle when their premiums escalate.
Loretta Andreasson has lived in the same Cape Cod ranch house in Harwich, Mass., for 30 years, and in all that time, a hurricane hasn’t come close to damaging her property. But in late 2006, her longtime insurer, Patrons Mutual Insurance Co., decided the risk was too great and dropped her policy. The company says it chose not to renew 844 policies in the Cape in 2006, based primarily on a 107% increase in their cost of reinsurance. Patrons acknowledges that it used a cat model from RMS in its decision-making.
With no other insurer willing to take her, Ms. Andreasson had no choice but to join the Massachusetts FAIR plan. Her premiums jumped to $1,178 a year from $492 over two years. “I should be able to retire,” says the 68-year-old dog groomer, who notes that a confluence of rising expenses have taxed her modest earnings of roughly $15,000 a year.