"Assessing risks fairly and sustainably"

Silvio Tschudi: We generate artificial hurricanes on the basis of historical data (that is, storm paths of previously observed hurricanes) and their intensity (hurricanes are divided into 5 categories according to wind velocity). This means that we conduct computer simulations of many artificial events that accurately reflect hurricane history yet have not yet actually occurred.

Later on, we can make statements about the intensity and frequency of future events on the basis of these artificial events. The intensity and frequency, in turn, form the basis for determining the anticipated damage. Thus, there is a correlation between wind velocity and the damage that can be expected, and we represent this correlation in a "vulnerability function," in which the damage naturally depends not only on wind velocity but also on whether many or only a few insured assets are located in a place. The end result of a risk analysis is information about the frequency or probability of a loss in an insured portfolio.

When an individual event, such as Irene, is currently taking place we monitor it continuously and once we know with a sufficient amount of certainty how this event will continue to develop, we can re-simulate it in our models, using past events that have already occurred. This gives us the opportunity to estimate potential damage relatively accurately. "Hurricane Gloria" from 1985 turned out to resemble Irene and was therefore suitable.

In addition to storm paths and intensity, storm surges associated with the hurricanes are also often evaluated, although they are much more difficult to represent.

What factors make this calculation so complicated?

Tschudi: As with any probability-based risk assessment, there are many sources of uncertainty. When modeling hurricanes, for example, the uncertainties lie in the storm surge component. In addition, it is difficult to calculate whether and to what extent the hurricane will gain or lose energy along the way and even what path it will take.

The "landfall" question, or in other words whether or not a hurricane will strike land, is always decisive. This naturally has an enormous impact on the level of damage.

There are also uncertainties involved in "translating" wind velocities into potential damage; unfortunately, this is true for all models.

There are other general uncertainties in the probability-based assessment of the consequences of storms. We talk about probabilities of a loss, i.e., uncertainties exist per se in the final statement. The question is only how far the modeled statements are removed from reality, and we tend to become constantly better and more realistic in our calculations.

Does climate change also play a role?

Tschudi: Absolutely. According to the latest research, the number of hurricanes is not necessarily increasing, although they are likely growing in intensity.

We can very effectively assess many factors involved in climate change. One such factor is the increase in surface water temperature, which supplies the energy for hurricanes. Hurricanes require a surface water temperature of 26 degrees Celsius or higher in order to be generated.

Other phenomena influenced by climate change, such as El Niño – which also have an effect on the frequency of hurricanes – are much more difficult to assess, and they involve greater uncertainty in the climate models.

What do these considerations mean for insurers and reinsurers? What risks can still be covered in the future?

Tschudi: Our risk models are aimed at evaluating potential risks due to natural disasters in a way that ensures the greatest possible protection. When it comes to protection, we must nevertheless take into account the uncertainties discussed above.

In addition to climate change, which continuously alters the conditions for future storms, the somewhat unclear set of data is a primary driver of the uncertainty. Often we don't know all relevant details that affect an insured property.

In concrete terms, this means that we are constantly reviewing our models and comparing them to actual events that have occurred. This is the only way to ensure that our risks are evaluated fairly and sustainably.

Silvio Tschudi

 
As with all content published on this site, these statements are subject to our Forward Looking Statement disclaimer.
Link to the disclaimer