Last year a student of mine, "http://sciencepolicy.colorado.edu/homepages/edouard_h/">Edouard von
Herberstein, wrote a fantastic master’s thesis on the use of
catastrophe models in insurance and reinsurance decision making. It is
hard to find on our WWW site, so it is worth highlighting. Here is the href =
"http://sciencepolicy.colorado.edu/homepages/edouard_h/hurricane_risk_pricing/">abstract:

Hurricane risk pricing, catastrophe models, and data quality: Why it
matters and what should be done about it?

Over the past 20 years, the colossal increase in computing power has
allowed computers to simulate very complex systems that require millions
of calculations and operations per second. Simulation software is
frequently used to forecast weather, exchange rates fluctuations, stock
price movement, or global climate. In most cases, computer simulation is
the only available tool generating forecasts from complex models. Assuming
that the use of more information in more complex simulators reduces
uncertainty, decision makers often incorporate these forecasts in their
decision processes. In some cases, decision makers give simulation tool
results a very large weight in their final decision.

Catastrophe models are a good illustration of very complex computer
simulations that largely drive business decisions in the insurance and
reinsurance industries. But just as in global climate models, the
sensitivity and uncertainty in catastrophe models should not be overlooked
when using model output in decision-making. In this project ENVS graduate
student "http://sciencepolicy.colorado.edu/homepages/edouard_h/">Edouard von
Herberstein proposes a method to assess the sensitivity of insurance pricing methods to data quality and questions whether these pricing techniques efficiently use the information in hurricane loss models.

See Edouard’s complete report "http://sciencepolicy.colorado.edu/homepages/edouard_h/hurricane_risk_pricing/final_report.pdf">here.