Medium-Term Hurricane Catastrophe Risk Models and Sensitivity to Sea Surface Temperatures

Climate oscillations and trends are a major source of uncertainty in dealing with hurricane risk. For decision makers a fundamental problem is to define the appropriate reference timeframe from which to evaluate future risk. A common practice in catastrophe loss models, a widely used tool in the insurance and reinsurance communities, is to build a risk model referring to the longest available time series of hurricane data. This approach provides a long-term historical perspective on hurricane risk, by integrating over the climate variability of the entire period for which observations are available. Although at first glance this approach seems logical, it neglects any relationship between climate variability and hurricane risk in time scales shorter than the entire historical record. It is therefore not clear that this approach provides the best assessment of the current hurricane risk. A more appealing, although admittedly more difficult approach, involves quantifying hurricane risk according to shorter or ‘medium-term' timescales that capture the interactions between hurricane activity and climate variability. Validation processes can then be used to demonstrate whether such a loss model is better able to predict hurricane losses than a long-term model. Such an approach has clear advantages, in that it integrates the capability to assess risk based on shorter time scales of the climate that influence hurricane activity. Here we present the latest research from RMS in developing such a ‘medium-term' view of risk and explore to what extent the skill to predict hurricane risk comes from the ability of the forecast system to capture changes in Atlantic sea surface temperature. Furthermore, we show some results on the usefulness of near term and decadal climate prediction to improve the statistical skill of hurricane risk