Maintaining a safety net: there's a growing trend in the use of predictive analytics as an answer to intensifying competition.

Fans of old movies may recall a dramatic scene in Cecil B. DeMille's 1952 epic The Greatest Show on Earth when the trapeze artist cuts his safety net and plunges to a crippling fall.

No serious insurer would intentionally remove a safety net that could safeguard profitability, but more than a few might fail to recognize the new safety nets emerging in the property/casualty business. Thanks to recent technological advances, insurers have access to strong safety nets that can save them from miscalculations which might plunge them and their companies into poor results or worse.

The development and use of predictive analytic tools is growing because of the convergence of improved technology and the industry's need for greater rating accuracy requiring greater quantification of risk.

More insurers are turning to the new complex technology of analytics to produce simple solutions to underwriting, rating and claims management challenges. Analytics incorporates advanced actuarial science, statistics, mathematics, econometrics, operations research, marketing data modeling and decision theory to produce simpler, but more powerful tools to discover relationships and predict outcomes. By linking financial, geographic and census data to loss records and employing new, more flexible actuarial methodologies, property/casualty professionals are using analytic modeling to achieve greater underwriting accuracy.

Two personal auto risks might look the same to Company A based on its classification plan. Both male drivers--living in the same ZIP code area, approximately the same age, and driving similar make and model sedans--might represent the same exposure to Company A and, accordingly, would be charged premiums based on the same rote.

But to Company B, those same two risks, when scored by an analytic model that incorporates nontraditional variables (such as years of driving experience, salvage history of the vehicle and traffic levels on the driver's street) with traditional variables of a class plan, might represent significantly different levels of risk requiring significantly different rates.

In the race to achieve sound and profitable underwriting without undermining its competitive edge, Company B gains a clear advantage from more accurate risk evaluation through analytic modeling.

Likewise, the advantages of using analytic modeling in underwriting small commercial risks--a class of business known as business owners policy or BOP--are significant when it comes to improving loss predictability. Fast and accurate scoring can mean the difference between insuring a good risk and taking a gamble.

But in the BOP business, where premiums average $1,500, carriers may not find it cost-effective to invest significant staff time and effort for painstaking hands-on underwriting. They know a key to making the sale is being able to respond quickly to an agent's request for a price quote. But in producing quick quotes, underwriting standards can suffer and insurers may take on business they'll regret.

To address this challenge, the Insurance Services Office recently introduced ISO Score, an analytic scoring model that predicts the likelihood of loss for individual small-business risks by assigning a score ranging from 100 to 1,000, with 1,000 representing the least likelihood of loss. The model is used to underwrite apartments, condominiums, small retail and wholesale outlets, offices, service establishments and fast-food restaurants, among other risks.

Besides personal auto and BOP, other promising lines for predictive analytic scoring include homeowners insurance and workers' compensation.

We are seeing a growing trend in the use of predictive analytics as one response to intensifying competition in the property/casualty market. Insurers are looking for better, quicker and less expensive ways of managing risk and making decisions.

An industry model created for carriers writing similar types of risk can produce significant economies of scale and eliminate the large start-up costs associated with modeling systems built by consultants. A model designed to serve the industry, yet still company-distinctive, requires little or no internal system development, can be reasonably priced and available for use right away.

A history of poor profitability has contributed to a dramatic decline in the number of private insurers serving the U.S. market, down from 1,272 in 1990 to 939 at the end of 2003. Survival and success depend on better execution. To grow premium while avoiding adverse selection, an increasing number of insurers will begin to utilize the performance safety net in analytic technologies.

Frank J. Coyne, a Best's Review columnist, is chairman, president and chief executive officer of ISO. He can be reached at [e-mail].