Assigning Risk to Targets of Terrorism - A Location Intelligent Approach
Editor-in-chief Joe Francica interviewed Simon
Sole, CEO of London-based Exclusive
Analysis, and Ben Hirsch, director of business development for
Insurance at Pitney Bowes MapInfo
about a collaboration between the two companies.
Joe Francica (JF): Please provide a brief overview of your criteria
for assessing risk by country or region. What percentage of those
criteria would you say are related to certain location-based risk
factors (environmental impact, susceptibility to natural disaster,
Simon Sole (SS): Exclusive Analysis' core activity is forecasting
violent and political risk globally. Within this purview, it is violent
risk that is primarily location-based. It is true to say that war,
terrorism or civil unrest always occur somewhere. When a bomb detonates
or a riot breaks out, it occurs at a known location, meaning the event
can be captured by a global Current Intelligence network and used
within our models. Likewise, when we advise clients on the risk of
terrorism, it is always possible to precisely define the location of
the risk. This is exactly shown by the TerrorRisk product that allows
us to precisely differentiate terrorism risk, right down to building
level. Conversely, it is far more difficult to relate political risk
(such as trade credit risk, nationalization, regime instability or
contract frustration) to location, as it clearly does not apply to a
JF: Explain briefly a typical model that you would develop that
would incorporate location-based risk factors.
Ben Hirsch (BH): TerrorRisk and Pitney Bowes MapInfo location
intelligence capabilities can be used in a variety of ways to analyze
terrorism risk exposure.
For instance, TerrorRisk's 3,700 global points of interest, with their
accompanying numeric Risk Values, can be added to MapInfo Professional
where they can be layered in with an insurance carrier's existing
policy portfolio in order to assess aggregate exposure to terrorism
loss. Policy accumulations can be calculated around the points (for
instance using rings or grids) and thematically mapped in order to
locate terrorism risk "hotspots,' identify potential areas to setup
exclusionary zones, or to flag regions where additional reinsurance
coverage will be necessary.
For any given location, multi-ring studies can be used to model various
loss scenarios that might result from terrorist actions at a given
location and quantify what the impact would be to the carrier across
its lines of business.
Another scenario would be to automatically load the TerrorRisk
intelligence into a Web-based underwriting application created using
Envinsa, or leveraging the Envinsa Online Services. The underwriter
could view any policy up for consideration in relation to the
TerrorRisk points (as well as any other policy risk factors of interest
- for example, historic weather perils) to better rate the policy.
If an insurance company has a better understanding of its potential for
loss in a given area, it is much further ahead of managing its risk
than it was without this insight. The carrier can also more clearly
demonstrate its terrorism risk exposure to rating agencies such as A.M.
Best, which evaluate the carrier's preparedness in their supplemental
JF: How does mapping/GIS technology support your model development?
BH: Geospatial analysis puts this layer of data to work to develop
insight for making sound business decisions, something that has become
known as location intelligence. Realizing the proximity to such
exposures is a consideration in the underwriting of an insurance policy
or the site selection for an asset such as a store or cell tower.
Understanding how much exposure already exists in close proximity to a
scored Point of Interest (investments such as mortgages, insurance
coverage, or inventory) allows better enterprise risk management
decision making, including business interruption planning, portfolio
risk management, as well as insurance/reinsurance decision making.
JF: Are you assigning risk to certain land-based features?
SS: The current iteration of TerrorRisk allows the client to
understand the terrorism risk at specific targets. The next iteration,
currently in development, assigns risk to entire cities, regions and
countries with a progressive granularity that allows street-level risk
differentiation in major cities.
JF: Are you assigning risk to weather or other random, transient
BH: Pitney Bowes MapInfo has developed its Risk Data Suite, which
contains historical weather and geological event data to help
underwriters, actuaries and risk managers to better assess exposure to
"natural" perils. Risk Data Suite comes with aggregate exposure grids
that can be used to help assign risk scores to properties and assets.
Subscribers can also use the historic data (alone or in conjunction
with other data they may have) to develop their own custom risk
exposure grids in MapInfo Professional or other spatial applications.
Using the grid scoring approach, analysts are able to easily return a
combined weighted score for a geocoded point. For instance,
policyholder location based on historical propensity and severity of
wind, hurricane, hail, lightning, flood, tornado, earthquake and other
layers of data can be used in underwriting a business decision. MapInfo
can also provide location intelligence solutions that incorporate
real-time weather data (converted into polygons with associated weather
severity ratings) and forecasts that can be used by insurance
companies, call centers and disaster response hubs to better prepare
for and respond to impending natural disasters. In insurance, both
claims departments and reinsurance, financial reporting activities
hinge on timely and accurate estimates of the potential impact of
JF: Are you assigning risk to man-made features like transportation,
SS: The TerrorRisk dataset includes those targets that we consider
at greatest risk of terrorism. Naturally, many of these potential
targets are buildings - for example, government offices, abortion
clinics, major mass transit hubs and so on. Additionally, targets that
are at risk of attack but are not discrete points, such as rail or
pipeline networks, present greater challenges of how to assign risk.
JF: Can you explain the interaction between spatial and non-spatial
phenomena and how you would visualize this type of interaction of risk
SS: Clearly there are area-applied risk metrics, for example
Chavez in Venezuela, or financial risks that are not geographically
located to a specific position. These types of risks cannot be directly
attributed to a particular target, but area factors do contribute to a
number of input indicators that affect target selection.
JF: Does the Country Risk Evaluation and Assessment Model (CREAM)
include a map interface? After clients use CREAM, are they provided
both a written and map-based report?
SS: CREAM is the industry-leading online facility. It models
violent and political risk globally across asset type and location to
provide actionable, intelligence-led forecasting for the effective
management of commercial risks worldwide. The combination of our
propriety intelligence methodology with our forecasting expertise sets
Exclusive Analysis apart and has enabled us to build an international
reputation as being at the forefront of strategic intelligence. It
includes both spatial and non-spatial features.
Written material is split into 'Assessments' that provide a topical
and/or thematic forecast of a particular issue in a given country and
'Risk Briefs' that provide a snapshot of the risk profile in each of
these risk classes for all countries. Both forms of analysis are
divided into risk classes and sectors to ensure a tailored service.
Metrics are accessible in the form of Risk Ratings. War, Terrorism,
Civil Unrest and Political Risks are given numerical ratings on 30-day,
one-year and three-year horizons that are designed to enable the
comparison of different risks within a country, the same risk over time
within country and the same risk between different countries. Spatial
tools are focused on high risk countries, where CREAM models risk at
specific asset locations using our innovative Interactive Country Risk
Maps. Individual country maps can be enlarged and show "risk clouds"
and level of risk generated by events in specific areas. The maps can
also be customized to show the scale of risk by asset type, or the risk
to key road and rail cargo routes.
Although the current joint offering of Exclusive Analysis and MapInfo,
TerrorRisk, is limited to scored Global Points of Interest with
continued updates and supporting advice, we have additional products
and services planned for introduction to the market, including the
Web-based CREAM services that we plan to offer as a Web service.
Frankly, we have been careful not to overwhelm the market, as it is
just now maturing to the whole prospect of developing location
intelligence around terrorism and political risk as a part of everyday
business. So, yes, CREAM includes mapping and we will be expanding
these offerings to meet market demand which we already see evolving.
JF: Are your clients finding mapping technology beneficial and what
elements do they see as helping them to best assess risk?
BH: In insurance, retail, telecommunication and banking we have
seen business people recognizing location intelligence as a required
competency. Exclusive Analysis' partnership with Pitney Bowes MapInfo
is augmenting the breadth of analyses that clients can employ in their
daily business decision making. Our clients have found value in being
able to go to one trusted location intelligence advisor that can
provide all the necessary ingredients for them to be able to deploy a
solution that meets their business needs. Pitney Bowes MapInfo can
provide not only the software and data they need and ensure that it
integrates together seamlessly, but also provide industry-specific
experience in solving similar problems for other clients.
Mapping and visualization business users understand relationships
between data in new and powerful ways. We've always known this, but
have not necessarily had easy means to incorporate consistent and
reliable geospatial analyses into existing business workflows. Web
services has been a major enabler of this change. Let's not overlook
the value of being able to use Web services to "feed" answers like "Is
this location proximate to a TerrorRisk point of interest?" back to a
system that will only involve a human if the answer is unacceptable and
requires intervention. This is something that is revolutionizing the
efficiency of decision making in the insurance industry.
JF: Would an actionable assessment of risk tell your client "where"
the risk factor is highest and how do you define the accuracy of
"where"? For example, can you assess the risk factor for the detonation
of an IED in downtown Baghdad within blocks? Square miles?
SS: The specificity and accuracy of the forecast to an extent
depends on the risk being assessed and the intelligence available. For
example, our Middle East analysts have identified a small number of
very specific military and energy installations as being at significant
risk in Bahrain - a country with an otherwise low terrorism risk
profile. Conversely, large areas of Baghdad share a similar risk
profile so the ability to differentiate IED risk on a street-by-street
level is limited. It is also true that modeling risk in high-threat
environments, such as Baghdad, necessitates different approaches to
those applied to London or Manhattan. In Iraq, for example, a high
event density (multiple attacks occurring per day) allows risk
differentiation to be based heavily on event data. In Manhattan, where
attacks are a rarity, a different approach to modeling is required; one
based more on forward-looking intelligence and indicators, such as
analyses of terrorist group intentions and capabilities.
JF: What is your intended future use or deployment of location-based
technology to help you model risk?
BH: While client demand and maturity will drive the product life
cycle, there are several ways in which we intend to integrate
location-based technology and modeling risk. One is to develop into the
emerging markets. We are also looking to make location the portal for
analysis, enabling users to search for intelligence and analysis
against specific locations, accessible via maps. Location-based
technology will let us relatively risk rate books of business; we aim
to provide differentiated treaty risks ratings. Finally, location
intelligence allows us to carry out advanced, asset-specific,
location-specific analysis to provide accurate Monte Carlo simulation
driven risk forecasts.