Terrorism Models Help, But Only Go So Far

Underwriters eager to avoid large concentrations of risk in case a terrorist strikes

Terrorism risk modeling might never reach the relative exactitude of its natural disaster counterpart, but the science is progressing despite most of the exposures inherently unquantifiable metrics. As the industry waits for Congress to extend the Terrorism Risk Insurance Act, such calculations take on new importance.

Two of the main vendors of such models are Boston-based AIR Worldwide and Risk Management Solutions in Newark, Calif.

Last October, RMS released a version of its U.S. Terrorism Risk Model that indicates a greater threat exists now than in 2003. “A record number of macro-attacks worldwide, a substantial increase in the number of planned attacks within the U.S., and increased activity in Islamic militant recruitment and rhetoric all lead us to believe that a risk of a major attack has increased,” said Gordon Woo, chief architect of the RMS model.

However, the expected severity of any such attack has lessened. “The chances of an attack involving a chemical, biological, radiological or nuclear weapon have decreased because U.S. and global defenses have improved the capability to detect such weapons, and the suspected Iraq weapons stockpile has not materialized,” according to Mr. Woo.

If a TRIA extension fails to materialize, the industry will still likely seek terrorism risk exclusions in each state, but by law they must cover the exposure for workers compensation policies. Thus, carriers will use exposure accumulation tools to ensure such risks are not too concentrated so as to result in unsustainable losses in the event of an attack on such a site.

Industry officials see progress on modeling terrorism riskup to a point. Indeed, despite the inherent inability of the models to predict the “when and where” of terrorism attacks, they still play an important role, according to Don Griffin, executive vice president for the Des Plaines, Ill.-based Property Casualty Insurers Association of America. “Companies will use the model because they still have to provide coverage in personal lines and workers comp,” he said. “It will change how companies underwrite and how they choose to exclude.”

Just how do modelers go about predicting terrorism risk? All for the most part rely on the hard data from years of military modeling on the impact of numerous kinds of weaponsboth conventional and nonconventionalalong with the insights of academics and other counterterrorism experts.

Peter Ulrich, managing director at RMS, said that with two World Trade Center attacks along with the Oklahoma City bombing, there is some data to work with. However, for the most part modelers rely on the expertise of academics who have studied Islamic attack patterns as well as the plans of schemes that did not come off.

Portfolio management remains the key service of terrorism risk modelers. “A company will come to us and say, We have written 400,000 terrorism policies, so what is our risk?” Mr. Ulrich noted. “Then when you have an underwriter in the field who has just written a new risk, you can figure out what it will do to your portfolio.”

As for the natural disaster analogy, Mr. Ulrich would compare Sept. 11 to Hurricane Andrew as seminal events in the two disciplines.

“There is certainly less scientific data like seismisity and soil data. I certainly hope we never have that much scientific data on terrorism, but in terms of acceptance and usage it is the same companies that have learned their lesson on catastrophe modeling that are now using terrorism modeling,” Mr. Ulrich explained.

If TRIA is allowed to expire, underwriters will have some pretty difficult decisions to make in nonexclusion states. “The models will definitely take on new importance if TRIA is not extended,” Mr. Ulrich said.

AIR modelers claim to have the first detailed terrorism lost estimation model to provide fully probabilistic loss costs. AIR Senior Manager Jack Seaquist said companies will use its modeling software throughout the year in the underwriting process to ensure that there is not too much exposure in any individual line.

Mr. Seaquist said that with input from AIRs team of counterterrorism experts, a “landmark database” with over 300,000 potential targets of attacksincluding commercial, industrial and educational siteshas been developed. “A subset of trophy targets carries a higher probability of greater risk,” he said.

Software developed for the U.S. Defense Department and the Federal Emergency Management Agency simulates various forms of attackincluding chemical and biologicalto measure their potential impact.

There will always be limits, however. “The models can tell you what will happen if there is a particular kind of bomb, but they cant tell you if it will be that bomb or when it will take place because of the human element,” Mr. Griffin said.

Reproduced from National Underwriter Edition, January 20, 2005. Copyright 2005 by The National Underwriter Company in the serial publication. All rights reserved.Copyright in this article as an independent work may be held by the author.

Special Reports /

Terrorism Models Help, But Only Go So Far

Related Stories

Recommended For You