Since last fall, Gordon Woo has spent his days poring over al Qaeda training manuals and surfing terrorism Web sites, hoping to uncover patterns that will help him predict the nature and frequency of future terrorist attacks.
But Dr. Woo doesn't work at the Defense Department. He consults for insurance companies, and is currently putting his mind to work on the problem of how to price insurance policies covering terrorist acts. Among his tools: game theory, the statistical approach made famous in the recent movie "A Beautiful Mind" about the life of mathematician John Forbes Nash Jr.
In a paper on the subject, Dr. Woo predicted that the frequency and severity of future attacks would be influenced by the structure of terrorist networks, which he compares to insect swarms and German U-boat fleets. This perspective helps "get under the skin of these groups and figure out what their strategies will be," he explains. Gathering enough information on terrorists' swarmlike behavior, the nature of past attacks and the goals of al Qaeda will provide parameters for models to price insurance, he adds, though such modeling is in its early stages and remains a formidable task.
Until now, insurance executives, whose companies will pay an estimated $50 billion in claims related to Sept. 11, have said terrorism is a largely uninsurable risk, and underwriters generally are excluding terrorism coverage from commercial-insurance policies as they renew. A handful of insurers have begun offering terrorism insurance as an expensive add-on to policies, but rates are high and the coverage terms narrow.
The seemingly incalculable risk is behind the industry's effort to get the government to create a backstop terrorism-insurance program that would pay most of the costs of such claims. Legislation has been stalled in Congress, in part because some congressmen are concerned about appearing to "bail out" the insurance industry, and prefer a program that imposes more obligations on insurers. But suddenly there are new signs of life for it. President Bush will give a high-profile speech Monday prodding Congress to act soon. To make the case that the absence of federal terror insurance is hurting the economy, the White House has lined up executives to talk about how their businesses have been hurt.
"Until two weeks ago, we thought the chances were 50-50 at best" for legislation, says Clifton Rodgers Jr., senior vice president of the Real Estate Roundtable, which is a leader of the recently formed Coalition to Insure Against Terrorism. He now puts the odds at 65%.
![[chart]](11.gif)
Most insurance companies are reluctant to offer coverage for terrorist events because they can't calculate the risk, and thus set the appropriate premiums. "The challenge is there's no type of consensus about what the probability of a terrorist event might be," notes Keith Buckley, managing director at Fitch Ratings, an insurance-ratings firm. "With natural catastrophes, they can look at the historical frequency. For man-made catastrophes, history doesn't repeat itself."
That is where game theory comes in. It tries to help explain how two or more adversaries would or should act during a conflict. While intrigued, not all academics are entirely convinced that it's applicable to terrorism. "To use game theory you have to have some well developed goal of the parties playing the game," says Tom Russell, an economics professor at Santa Clara University in Santa Clara, Calif. "When you talk about suicide bombers, it's not completely clear what the goal is."
Dr. Woo, a consultant with Risk Management Solutions in London, concedes that it is too soon to begin pricing terrorism risks. Still, he and others maintain that game theory could suit the needs of insurers trying to assess such risks because they need to better understand what buildings terrorists might target, and how likely attacks are to succeed given the defensive strategies of the government and business owners.
A simple example: Two duelists with pistols stand 50 yards apart and walk toward one another. "When do you fire the shot?" asks Dr. Woo. "Game theory helps answer that question. But the nature of the game changes if, like these terrorists, you're into suicide. If you believe you're going to paradise if you die, you tend to want to fire later."
One surprising result of early game-theory modeling in the insurance sector: The statistical likelihood of monetary damage caused by terrorists to the most desirable targets -- say, large landmark buildings in New York -- is the same as more mundane targets. The reason is that the more desirable the target, the more likely it is to be well defended. That conclusion, drawn by game-theory enthusiast John Major, an actuary and senior vice president at reinsurance broker Guy Carpenter, a unit of insurance brokerage company Marsh & McLennan Cos., is much different than the ones drawn by insurance companies since the attacks. Thus far, insurers have charged huge premiums to provide terrorism coverage for likely target properties, and many property owners have chosen to go without the coverage.
Many insurers are convinced that terrorism insurance could be profitable if the risk can be quantified. Insurers, for example, have collected more than $500 million in premiums from airlines for war and terrorism coverage since Sept. 11. Before then, most insurers provided such coverage, which airlines are required to carry in order to fly into most airports, as part of a basic package. The rates charged on the coverage -- which U.S. airlines have declined to buy because of its price, instead turning to the government for coverage -- is based on rough estimates that another terrorism-related loss would occur in approximately 18 months.
But such estimates are little more than guesses, and insurers have thus far offered little coverage for other types of losses. In November, Dr. Woo, a mathematician trained at Cambridge University and Massachusetts Institute of Technology and author of the book, "The Mathematics of Natural Catastrophes," set out to build models that would allow insurers to broaden their coverage.
One of his first steps was to read research reports on terrorist threats produced by Rand Corp., the same nonprofit research organization where, in the 1950s, Mr. Nash used some of his game-theory ideas to predict how the Soviet Union would deploy its nuclear arsenal. In addition, Dr. Woo has explored Internet sites like terrorism.com and literature like an al Qaeda training manual posted on the Justice Department's Web site. "They actually specify targets," Dr. Woo says of the manual. Among goals listed in the manual: "Blasting and destroying the places of amusement, immorality, and sin. … Blasting and destroying embassies and attacking vital economic centers."
Dr. Woo presented his first paper on the subject in early February in Cambridge, Mass., at a meeting of the National Bureau of Economic Research Insurance Group. He said old methods of estimating insurance losses from earthquakes and hurricanes would be inadequate for evaluating terrorism risks. A new model using the input of experts such as government security agencies is needed to predict the frequency and severity of future terrorism-related losses, he added.
Guy Carpenter's Mr. Major followed Dr. Woo's presentation with a critique, in which he brought game theory into the discussion. His conclusion: Game theory has "the potential to offer useful insights to the insurance profession."
Insurance companies that might use game theory to price policies remain skeptical of its applications to their work, and some still insist a government backstop is the best way to cover future claims from terrorist attacks. Gordon Stewart, president of the Insurance Information Institute, a trade group, says insurers need government help to pay terrorism losses because the government controls the intelligence networks on terrorist activity, as well as the armed forces responsible for rooting out the activity.
But Dr. Woo says the insurers need to make an effort to catch up with the government in understanding terrorists. "What's happened so far is that the government has done all the running," he says. "Typically the insurers are late in addressing these kinds of problems. There's no reason, though, why they shouldn't be much better prepared."