Following Congressional failure to renew the U.S. terrorism insurance backstop, industry professionals agree that terrorism exposures can modeled and may even be an attractive risk for securitization if traditional insurance and reinsurance players pull back from the market.
But with an almost infinite possible maximum loss — and a limited historical data set — any firms jumping into terrorism risk could face significant obstacles for success.
“There are solutions that address the risks that were built over the last decade, but they address the limited events that have been experienced,” says Henry Willis, director of RAND Corp’s Homeland Security and Defense Center. “They don’t address the terrorism we haven’t seen, and that’s the issue.”
On Wednesday, the Senate failed to garner enough backing to extend the Terrorism Risk Insurance Act (TRIA), meaning the legislation will sunset on Dec. 31, 2014. Passed in 2002, TRIA made the federal government the insurer of last resort for terrorism risk, and the most recent version of the legislation would have extended the program for six years. The bill would have also raised the minimum trigger to $200 million from $100 million and increased insurers’ share of the cost of losses from future catastrophic terrorist attacks to $37.5 billion.
Catastrophe modeling firms have been peppered with terrorism risk questions in the months leading up to the legislation’s eventual defeat, says Chris Folkman, director of project management at RMS.
“There’s definitely been more interest in the details about terrorism modeling,” Follkman says. “That makes sense because — even in the best of outcomes for TRIA renewal — the insurance industry is going to be on the hook for billions of dollars of more terrorism risk. So they are much more interested in understanding the nuances of the model and how that impacts their balance sheets.”
“If TRIA expires, terror risk models can provide invaluable insight for insurers looking to evaluate and mitigate the resulting impact,” said Alissa Legenza, risk consultant at AIR Worldwide in an email statement. “Terrorism remains a highly dynamic peril, capable of causing catastrophic insurance losses across multiple lines of business, without any warning. Coupling this threat with the uncertain fate of the TRIA renewal, insurers and reinsurers are increasingly turning to sophisticated catastrophe modeling tools designed to quantify and manage the risk of losses from future terrorist attacks.”
Folkman adds that several insurance-linked securities (ILS) shops have approached the firm, interested in how the models operate and whether it could fit into their product lineup in the event traditional players begin to withdraw.
“We’ve gotten renewed interest from ILS in understanding the dimensions of terrorism risk and leveraging that into opportunities. I think we’ll see terrorism risk securitized in some form of alternative capital deal,“ he explains.
But despite the interest in the models, there are significant hurdles to measure terrorism risk accurately insure or securitize it, says RAND’s Willis. “We do have an understanding of frequency and the implications of conventional attacks, like the Boston Marathon bombing,” he says. “But how do you measure the likelihood and losses of a nuclear and biological attack? How do you measure and insurance against small, but sustained attacks over a period of time?”
According to Folkman, the basics of using a terrorism model are very similar in terms of the user to the RMS nat cat model. For example, the insurer uploads its data, which includes information about their properties and human lives at risk. In terms of hazard generation — unlike earthquake ground shaking or wind for hurricane’s — the hazard generation for terrorism is built around variables like the blast wave from a detonation.
“From there we look at vulnerability assessment and then we go to the damages,” Folkman says. “The damage assessment is very similar – what might that cost in terms of replacement cost.”
The key differences between terrorism modeling and nat cat, according to RMS’ Folkman:
- Terrorism events are very localized in nature, usually within a 400-meter radius. “The validity of address data in being able to pinpoint exactly where a person or building becomes very important.”
- Terrorism considers human exposure. “We model workers compensation loss. Buildings are fixed but human beings move around all day. There is a lot of uncertainty around the spatial distribution of people.”
- Attack multiplicity . “Multiple synchronous attacks occur in one coordinated terrorism event. You saw that with 9/11. You see it with car bombs. “
Despite the advances that modelers like RMS and AIR have made, RAND’s Willis remains skeptical.
“We have learned more [about modeling terrorism], but in many ways we still don’t know enough,” Willis says.