Hurricane predictions have little impact on preparedness, policies

Seasonal hurricane predictions are interesting but have little practical impact on how risk managers, insurers, catastrophe modelers and others go about their business, observers say.

This is largely because of uncertainty surrounding hurricane predictions, which are based on a complex yet still-evolving science (see story, page 28).

Risk managers and others need look only at the generally inaccurate predictions of a heavy hurricane season in 2006 to see they should not necessarily be relied upon, observers say.

Furthermore, these predictions usually are made too late to introduce effective loss prevention programs or to be reflected in that year's insurance or reinsurance policies.

Observers point out that--whatever the total number predicted--it takes only one hurricane to cause considerable damage.

Even so, there is general agreement that hurricane predictions have played a valuable, if less quantifiable, role in creating greater awareness of such risk.

Meanwhile, experts say the Atlantic Basin is in an active hurricane cycle that began in 1995 and could last another decade or two.

According to the Fort Collins, Colo.-based Tropical Meteorology Project at Colorado State University, the average of 1.5 major Atlantic hurricanes per year during the 25-year period ending in 1994 increased to 3.9 during the 12-year period ending in 2006.

The project has predicted that 17 named storms will form in the Atlantic Basin during the current hurricane season that runs through Nov. 30, of which five a projected to become major storms.

However, observers say the predictions are simply are unreliable for advance planning.

"We don't find them that useful," said Ann Brown, risk manager for New Hanover County, N.C., based in Wilmington. "Last year, they projected such a massive amount of hurricanes to come up the Eastern seaboard and we didn't have any major storms at all."

Lance Ewing, vp-risk management in Memphis, Tenn., for Harrah's Entertainment Inc., said none of the predictions warned of either Hurricanes Katrina or Rita in 2005, "so I take them with a little grain of salt."

"It is extremely difficult to give accurate weather forecasts beyond 10 days even of normal weather," let alone accurately predict hurricanes five to six months in advance, said Warren Ison, New York-based senior vp for Willis Re Inc.

Mark Saunders, head of seasonal forecasting and meteorological hazards at the Benfield Hazard Research Center at University College London and also lead scientist at London-based forecasting consortium Tropical Storm Risk, said a study of the forecasts issued since 1984 indicates it is not until early May that the forecasters have a relatively precise idea of how many intense storms there will be, and early July before they can accurately predict how many will make landfall.

Bill Steiner, managing director at reinsurance intermediary Guy Carpenter & Co. Inc. in New York, said predictions of CSU Meteorologist William Gray made six months in advance of hurricane season are "really of little value" in forming the catastrophe models used by underwriters "because the models are based on probabilities," not so much on how many storms will occur over the next year as "looking ahead in increments of a few years."

"It's much more difficult to make a precise, single-year estimate even when you have information up to the very start of the season," said Bob Ward, London-based director of the global science network at Risk Management Solutions Inc., a Newark, Calif.-based modeling firm. "Our clients don't ask us for seasonal forecasts," although such forecasts can be helpful in gauging whether a busy season is ahead because of the information the firm provides to insurers during and after major catastrophes, he said.

Furthermore, the forecasts are less useful in predicting which hurricanes will make landfall than they are in predicting a season's total number. "An awful lot of hurricanes die before they make it into land," and even then, there is a two-thirds chance they will skip major population centers, said Thomas Larsen, senior vp of Oakland, Calif.-based modeling firm EQECAT Inc. "Wind doesn't cause a loss," said Mr. Larsen. "It's blowing buildings down" that causes a loss, he said.

Observers also say the predictions neither immediately nor directly impact premiums. Past experience is the basis for insurers setting rates, said Eileen Gardner, risk manager of Brunswick County, N.C., based in Bolivia.

Noting it is not until July that hurricane forecasts are very good, Steve Smith, senior vp of ReAdvisory, the analytical arm of reinsurance intermediary Carvill Group in London, said, "The problem with that in terms of reinsurance buying, is it's all over" by then.

"Insurance is very illiquid," said Mr. Larsen. Most contracts take effect in January, June or July and once the contract is set, "There's not a lot of flexibility," he said. You "can't trade reinsurance contracts," regardless of whether the risk changes.

Insurers and reinsurers are interested in insuring against the one in 100-year loss, said Eric Brosius, senior vp and manager of reinsurance for Liberty Mutual Insurance Co. in Boston, which buys and sells reinsurance. So hurricane forecasts "make us feel better or worse about the decisions we make, but they don't have a large, objective effect on our decision-making," he said.

Some observers describe the predictions as just one tool. "They are a piece of how we at Harrah's prepare" for a catastrophe, said Mr. Ewing.

Claire Wilkinson, vp-global issues, at the New York-based Insurance Information Institute, said hurricane predictions "are one of the tools and one of the pieces of information that insurers can use as they look ahead." How useful they are "would depend company by company, according to what type of business you're writing and your portfolio."

But while they can be a useful part of the underwriting process, "it's important to remember that forecasts are just that" Ms. Wilkinson said. "They don't give you an absolute level of risk."

Nevertheless, long-range hurricane predictions have accomplished "raising everyone's consciousness and understanding about hurricanes," said Gary Kerney, vp, property claim services, at Jersey City, N.J.-based Insurance Services Office Inc.

Work by CSU's Mr. Gray and others has spurred research in hurricanes that predates the 120 years or so for which records have been kept. Research, for instance, has indicated that several severe hurricanes have hit the Northeast United States over the past 500 years, said Mr. Kerney.

"I think that people in the scientific community and folks in the insurance industry now have a better understanding of what the risk is because now we can see beyond the 1880 record, and now we understand there is a probability of significant and costly occurrence happening within the foreseeable future," Mr. Kerney said. Preparation in response to this awareness "makes the recovery that much easier," he said.