Hurricane risk pricing, catastrophe models, and data quality

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 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.


Click here [PDF] to view the Final Report