Author's personal copy
Statistical Methodology 7 (2010) 574–576
Contents lists available at ScienceDirect
Statistical Methodology
journal homepage: www.elsevier.com/locate/stamet
Discussion paper
Discussion on ‘‘Public Hurricane Loss Evaluation Models:
Predicting losses of residential structures in the state of
Florida’’ by S. Hamid et al.
Thomas H. Jagger
*
, James B. Elsner
Department of Geography, Florida State University, Tallahassee, FL 32306, United States
1. Introduction
The paper outlines the science and engineering approach used in developing the Florida Public
Hurricane Loss Model (FPHLM). Loss models are important because a purely actuarial approach to
projecting the future risk of large losses is limited by the amount of data on past damage amounts.
A public model is important because it provides an independent check on private model results,
which are used to set residential and commercial insurance premiums. We are happy to have this
opportunity to congratulate the authors on the development of this public loss model and on their
exposition of its various components. The authors have provided an important service to the public
and risk management community and the work will undoubtedly stimulate additional research on
various aspects of quantifying the expected losses from hurricanes at landfall.
2. Model components
The model consists of three components: (1) Tropical storm frequency and intensity (meteorol-
ogy); (2) Vulnerability of building construction type (vulnerability); and (3) Actuarial loss distribu-
tions (actuarial). To our knowledge this is the standard approach used by most risk modelers. In the
meteorology component, a 50K-year catalog of synthetic hurricanes is generated that mirrors histor-
ical hurricane activity in and around Florida. Synthetic hurricanes are generated using frequency and
intensity distributions derived from past hurricanes and a numerical model that provides near-ground
level wind conditions locally from equations that describe atmospheric forces inside a hurricane. In
the vulnerability component, a distribution of proportional loss is estimated for each combination
of building construction type using various wind speeds and wind directions. In the actuarial com-
ponent, synthetic hurricanes together with the vulnerability distributions are used to generate loss
distributions from a portfolio of insured properties. Expected annual losses are estimated for policies
at a spatial resolution of interest (at least down to the zip code level) in an insurance portfolio.
DOI of original article: 10.1016/j.stamet.2010.02.004.
*
Corresponding author. Tel.: +1 303 759 9776; fax: +1 850 644 5913.
E-mail address: tjagger@blarg.net (T.H. Jagger).
1572-3127/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.stamet.2010.06.003