Terrorism, political violence and other malicious acts differ from other insurable perils in that they are dynamic and adaptive; there is human intent driving the probability and willing the impacts to change. Historic incidents show past trends, but those are only partial drivers of future trends as terrorists adapt their tactics in response to past effectiveness and current circumstances. Additionally, the frequency of past successful attacks does not reveal the entirety of historic threat: unfulfilled, foiled, failed, redirected attacks, or those with unintended consequences, according to Stephen Hudson, Head of Terrorism, Global Strategic Advisory, Guy Carpenter.
There are further challenges in relying on historic frequencies as a proxy for future probabilities. Short return period events can be highly volatile, clustering one year and not the next, whereas long return period events have few or no data points to inform frequency.
These challenges are well understood, but there has often been a defeatist attitude in the insurance industry that these challenges are too difficult to address. The structure of a catastrophe model for natural perils is applicable to terrorism and political violence. “Modules are required for hazard, exposure, vulnerability and damage and financial impact. The challenge in creating a terrorism risk model is achieving sufficient precision and refinement of each module to make the model credible – particularly the hazard module, because of the changeable and dynamic nature of the peril,” adds Mark Weatherhead, Head of Model Development – International, Guy Carpenter.
Guy Carpenter’s proprietary terrorism model, Sunstone™, adds to the firm’s reputation for innovation. “The model contains over one million global targets, impact curves for 40 attack types, probability matrices for eight different regions, and a stochastic model that integrates these elements. The University of Maryland’s Global Terrorism Database of historic incidents is also included,” Hudson explains. “Unlike other models, Sunstone is truly global and reflects the full spectrum of attack types from large catastrophic events to lower level, more frequent events. As a result, Sunstone consistently provides a more credible model of potential losses. While Sunstone is an effective model, we are not standing still and are continually developing the model to better represent threat, vulnerability and impact.”
Weatherhead continues: “Vulnerability is the most absent element of terrorism risk modeling in the insurance sector, which in this context, is the structures, people and processes most vulnerable to attack – data that would ideally be provided by insureds alongside their schedule of values. If this does not occur, the alternative is for the insurance industry to research and inject the information. Clearly, the quantity of assets being managed highlights the scale of the challenge.”
Additionally, challenges arise in precisely assessing the impact of terrorist and political violence across a large portfolio. “Concentric ring modeling is a reasonable simplification of the effects of blast in a flat, open field-type location,” says Hudson. “But insurance exposures rarely exist as such, and most often occur in dense urban environments where an explosive blast may be confined by surrounding buildings, reflected off their surfaces and channeled down the man-made canyons of city blocks, shielded by intervening buildings.”
Guy Carpenter has responded to the challenge of blast impacts in dense urban environments by developing 3D cityscape blast modeling. “This capability,” explains Hudson, “provides a detailed representation of several square miles of central business districts.
It is able to complete computational fluid dynamic (CFD) modeling and simplified ray-trace analysis, then accurately assess the engineering response of the affected buildings. Guy Carpenter offers market-leading visualization at unparalleled modeling speeds, using software validated by U.S. Government testing. This approach can calculate losses that may be 35 to 40 percent lower than the concentric ring technique, releasing significant capacity and more cost-effective reinsurance to our clients. The 3D cityscape also provides the environment for accurate modeling of crowd flows and chemical plumes.”
Guy Carpenter is working to integrate a variable resolution approach to impact assessment by employing sophisticated CFD modeling at key areas of risk within a portfolio with more simplistic ray-trace and concentric ring techniques, elsewhere. Weatherhead adds: “This approach enables Guy Carpenter to deliver targeted, sophisticated and efficient exposure management to our clients.
This variable resolution approach is also being applied to threat and vulnerability assessment, applying greater effort where market exposure is greatest, in order to achieve greater certainty and precision in these areas. Also, where it is not possible to measure factors directly, correlatory factors can indicate the trend.”