Unknown may resist analysis.

The industry is keen to model terrorism risk, but such attempts to second-guess terrorists may never be very successful, says John Sanders.

Assessing terrorist risk has become a hot topic in the wake of 11
September 2001. Risk modellers and brokers have stepped into the knowledge
breach with products, but their accuracy and value remain untested.

Towards the end of last year, three risk modelling companies - Applied
Insurance Research (AIR), Eqecat and Risk Management Solutions (RMS) -
launched products to help (re)insurers quantify, price and manage their
exposure to terrorism. The catalyst was the 2002 US Terrorism Risk
Insurance Act, under which primary insurers must offer terrorism coverage
on commercial lines.

The models are currently limited to terrorist risk in the US and focus on
prestigious targets, particularly in major cities, and military and
nuclear installations. AIR, for example, has developed a landmark database
covering 300,000 locations, while Eqecat encompasses over 10 million
events and hundreds of thousands of high-probability targets.

Over time, the modelling may cover Europe and Asia, but that will depend
on demand from insurers and its success in the US.

Although very precise, the stochastic modelling of terrorism, which looks
at the probability of an attack and its impact on many targets, is

Samir Shah, managing consultant at consultancy Tillinghast-Towers Perrin
in the US, believes it may make more sense for smaller insurers to use an
analytical approach and "build in some sort of buffer in the pricing for
that modelling uncertainty". Nonetheless, RMS says interest in its model
is strong and that it has already worked with around 30 companies.

Brokers have a bash

(Re)insurance brokers Aon and Benfield separately announced developments
in terrorism modelling in February. Benfield launched an extranet system
for monitoring concentrations of terrorism risk in property and casualty
portfolios. The system, Expect (EXPosure Evaluation and Control Tool),
uses postal codes and their proximity to each other to estimate peak

Benfield's Paul Maitland says the system, unlike others, does not try to
model the likelihood of terrorist attack.

Meanwhile, Aon subsidiary Aon Re has completed its first multi-line risk
analysis of a combined portfolio of life, accident and property terrorism
and earthquake catastrophe risks for a major insurer. Its analysis
included an evaluation of the risks to the insurer's property

"Being excluded to date from the (US) Terrorism Risk Insurance Act leaves
the life insurance community saddled with considerable exposure," says
Roger Smith, managing director of Aon Re's accident and health reinsurance

Dual benefit

The main benefits of terrorism modelling are exposure control and pricing
information. The World Trade Centre attack showed that a terrorist act is
likely to be much more concentrated than a hurricane; it resulted in huge
losses on life, liability, business interruption, workers' compensation
and fine art lines as well as the property losses associated with

Apart from encouraging a reduction in risk concentrations, quantifying the
risk can change insurers' attitudes, says Peter Ulrich, managing director
of the enterprise risk management business of RMS: "Helping companies
understand the risk certainly doesn't decrease it. But it may make them a
little more willing to be out in the marketplace once they have the tools
to help them manage accumulations around key targets and assess what the
magnitude of risk is."

The data provided by the models is also crucial for pricing. "If you can
model the frequency and severity of terrorist attacks on different
targets, then you can convert that into what the losses would be
associated with those. If you can estimate the loss, then you can price
the insurance," says Mr Shah.

Terrorism modelling draws heavily on past catastrophe modelling and less
well-known areas such as riot modelling. Modelling the damage and the
insured losses is relatively straightforward. Hurricanes provide a useful
guide to insured losses, while organisations such as the Rand Corporation
and Eqecat's parent ABS Consulting have been analysing bomb and blast data
for 50 years or more.

Despite the similarities to catastrophe modelling, fundamental differences
do exist: far more natural catastrophe data is available; hurricanes and
earthquakes follow patterns and have physical limits; and, most
importantly, terrorist groups are very unpredictable.

As Jack Seaquist, terrorism model programme manager at AIR, puts it, the
hardest part is predicting where, when and how terrorists will attack.

"We can't use historical data because we're in a different geo-political
climate than in the past," he says. "And the frequency and severity of
attacks certainly have varied over time."

Mr Shah agrees that second-guessing the terrorists is the biggest

"The way terrorists behave is that if they think you can find a pattern,
they will try to do something different. So, it is hard to forecast," he
says. "It is very difficult to establish patterns."

In an attempt to overcome this problem, the modelling companies rely
heavily on panels of experts. These include people with backgrounds in the
US in the Department of Defence, the Central Intelligence Agency, the
Federal Bureau of Investigation and the Department of Homeland

On the technical side, the companies draw on organisations such as Jane's
and the Rand Corporation. RMS also taps into expertise at the University
of St Andrews in Scotland, which has some of the world's leading
authorities on Islamic militants.

The modellers also use game theory and other mechanisms to try to work out
what terrorists might do next. "The underlying peril is dynamic. You can't
just run an analysis and put it on the shelf," says Mr Ulrich of RMS. "The
peril will change over time. That's why we use a game theory approach to
modelling to take these factors into account." AIR uses a system of threat
assessment to rank the likelihood of different kinds of attack in
different locations.

Nonetheless, terrorism modelling is still in its infancy. Products have
been developed quickly and the modellers are now working to improve their
initial offerings.

Modelling natural catastrophes has come a long way in the last 15 years,
but some doubt whether modelling terrorism will ever match that

"I think part of it will get better," says Mr Shah, "but I think that the
nature of terrorism risk, with the secret threats and secret defences, is
difficult to model."
Click here for a
Špdf of this story
  • LinkedIn  
  • Save this article
  • Print this page  

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact [email protected] or view our subscription options here: http://subscriptions.postonline.co.uk/subscribe

You are currently unable to copy this content. Please contact [email protected] to find out more.

You need to sign in to use this feature. If you don’t have an Insurance Post account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: