Extended Abstract (16K)11C.4
Insurance sector interests in best-track data sets
Richard J. Murnane, Bermuda Biological Station for Research, Garrett Park, MD
Hurricanes result in uncalculated amounts of human suffering and caused US insurers to lose an average of $3 billion per year between 1989 and 1999. Such large losses make hurricanes striking the Gulf and East coasts of the United States one of the biggest risks facing property catastrophe insurers and reinsurers (collectively called (re)insurers). (Re)insurers estimate their potential exposure to hurricane losses using commercial and in-house proprietary risk models. This presentation will briefly explain how best-track data sets are the foundation for these hurricane loss estimates, describe how best-track data are used by (re)insurers, and discuss reinsurer's interest in developing the "next generation" of best-track data sets.
There are three different components to the risk models used by (re)insurers: a hazard component, a damage component, and a loss component. The hazard component provides the probabilities for events associated with hazards such as hurricanes, tornadoes, and earthquakes. The probabilities are derived from climatologies and historical records and the best-track data sets are the basis for most of the hurricane hazard components. Most (re)insurers do not use the best-track data directly, rather they use risk models developed by engineers and scientists employed by commercial risk modeling companies.
Tropical cyclone activity varies as a function of climate, however, almost all risk models base their loss probability estimates on climatological averages. The historical record is generally too short to provide reliable estimates of how tropical cyclone activity and landfall probabilities vary as a function of climate. However, some reinsurers are interested in new ideas for generating "synthetic data sets" that could provide better information on how landfall probabilities vary as a function of climate.
Session 11C, Tropical Cyclone Best Track
Wednesday, 1 May 2002, 4:00 PM-5:30 PM
Previous paper Next paper