For good reason, the industry has focused its catastrophe management activities on natural hazards.
Since 1970, 38 of the top 40 most costly CAT losses have resulted from natural hazards. Since 1989, global insurers and reinsurers have suffered multi-billion dollar losses from no fewer than 20 earthquakes, windstorms, tornadoes, hailstorms, winter-storms, fires, and floods. Moreover, there is a credible body of data and science that suggest that future events of greater magnitude than those that have recently occurred are not only likely, but also inevitable. RMS models suggest that the global
P&C industry can expect an average of $20 billion in natural catastrophe losses per year, including the possibility for single event losses in excess of $100 billion.
However, as a man-made catastrophe, the WTC attack suggests that the industry now broaden its view beyond natural hazards. Even prior to the September 11 terrorist attacks, man-made events such as industrial accidents, aviation losses, riots, fire, and explosion have contributed 20% of the total
CAT losses over the past five years. In addition, some of the worst man-made disasters such as the
Exxon Valdez, Bhopal, and Chernobyl could have cost the industry billions if they had occurred under slightly different circumstances. Even events on the margin of our contemporary history warrant a second look, such as the Great Chicago Fire of 1871, which prior to September 11 seemed irrelevant to today’s circumstances.
Incorporating terrorist attacks and other man-made super-CATs into an overall understanding of catastrophe risk is a difficult proposition. Setting aside the task of estimating probabilities, the
‘footprints’ of loss are disparate and, unlike natural hazards, difficult to constrain. Some events, such as the WTC attack or the September 21, 2001 AZF fertilizer plant explosion in Toulouse, France2, are
‘vertical’ in nature with geographically distinct and highly concentrated damage patterns. Other events, such as biological or chemical hazards (accidental or otherwise), or IT-network disruptions
(e.g. computer viruses), are ‘horizontal’ in nature, with more diffuse, often geographically discontinuous, impacts. Despite these challenges, it is prudent and possible to develop a suite of catastrophe scenarios, or ‘stress tests’, which can be modeled to developjfw