The Trouble with Terrorism Risk Models
Why studying the past cannot always stop the future
All this makes for an odd paradox. Only a few days before May’s announcement, modeling firm RMS managed to capture a few headlines with an update of its Probabilistic Terrorism Model, claiming there’d been a “decline in the frequency of macro-terrorism attacks” (major property damage, more than 20 fatalities and 50 injuries). It also claimed that large-scale attacks are down because of successful counter-terrorism measures and a weakened al-Qaida. The problem is that, if you look deeper, the update seems to raise more questions than it answers.
For instance, RMS says, “Worldwide, the frequency of macroterrorism attacks is 85 percent lower in 2014 than in 2006…” That’s great, but as they say with mutual fund disclosures, past performance is not an indication of future results.
Chris Folkman, director of model product management at RMS, concedes that the “global threat landscape is very fluid. What we try to do is just provide a data-driven view of the risk.” With that fluidity, it’s difficult to see how anyone could underwrite a terrorism scenario in Vancouver, when the band of psychopaths formerly known as ISIS weren’t even in the news half a year ago. Folkman can talk about targets, weapon selections, economic value and peak occupancy rates of buildings in the RMS database, and how it goes “about building an event catalogue that is large enough that it would capture the most likely forms of terrorism to be perpetrated in the cities that we cover”—but that doesn’t answer the question.
If you were a tourist in Bali in 2002, you would have found nothing, in any data set, to warn you about a bomb going off outside the Sari Club. “I think you could have said that about a lot of catastrophic events,” countered Folkman, who, a day after the recent 6.0 earthquake in California, applied some curious logic. “Not just terrorism, but the people in Napa yesterday never imagined, you know, that another 6.0 earthquake would occur…”
It’s not a very good analogy, considering that those in the Napa Valley live in one of the most unstable fault zones in the world, so they can’t be that unimaginative, and there’s an entire branch of science devoted to predicting a phenomenon of nature—so much so that UC Berkeley’s early warning system gave residents a 10-second heads-up (and they’re promising to do better next time).
“The best that we can do is assess the risk landscape as it is today,” insists Folkman. He says that RMS does “re-calibrate our model, including the frequency, as often as necessary to address a changing risk landscape, and next year, we are probably going to take a look at a number of other areas of the model and get them up to date as well.”
And that’s fine, but threats like Boko Haram and the Islamic State are quickly evolving, a point Folkman readily concedes— that these groups could soon be international threats. It seems obvious that both the U.S. and U.K. authorities already think of them as such. So today’s model could be made obsolete by tomorrow’s headlines.
Copyright 2014 Rogers Publishing Ltd. This article first appeared in the September 2014 edition of Canadian Insurance Top Broker magazine