Authors

C. C. Watson;M. E. Johnson

Comments

Authors: contact us about adding a copy of your work at STARS@ucf.edu

Abbreviated Journal Title

Bull. Amer. Meteorol. Soc.

Keywords

WIND; Meteorology & Atmospheric Sciences

Abstract

The results of hurricane loss models are used regularly for multibillion dollar decisions in the insurance and financial services industries. These models are proprietary, and this "black box" nature hinders analysis. The proprietary models produce a wide range of results, often producing loss costs that differ by a ratio of three to one or more. In a study for the state of North Carolina, 324 combinations of loss models were analyzed, based on a combination of nine wind models, four surface friction models, and nine damage models drawn from the published literature in insurance, engineering, and meteorology. These combinations were tested against reported losses from Hurricanes Hugo and Andrew as reported by a major insurance company, as well as storm total losses for additional storms. Annual loss costs were then computed using these 324 combinations of models for both North Carolina and Florida, and compared with publicly available proprietary model results in Florida. The wide range of resulting loss costs for open, scientifically defensible models that perform well against observed losses mirrors the wide range of loss costs computed by the proprietary models currently in use. This outcome may be discouraging for governmental and corporate decision makers relying on this data for policy and investment guidance (due to the high variability across model results), but it also provides guidance for the efforts of future investigations to improve loss models. Although hurricane loss models are true multidisciplinary efforts, involving meteorology, engineering, statistics, and actuarial sciences, the field of meteorology offers the most promising opportunities for improvement of the state of the art.

Journal Title

Bulletin of the American Meteorological Society

Volume

85

Issue/Number

11

Publication Date

1-1-2004

Document Type

Article

Language

English

First Page

1713

Last Page

+

WOS Identifier

WOS:000225557200012

ISSN

0003-0007

Download

Share

COinS