The annual cost of hurricanes and big storms could increase by two-thirds over the next seven decades (to USD 27 billion, GBP 15 billion) if CO2 emissions double, insurance industry representatives have warned. Even if emissions are significantly curbed, hurricanes remain a damaging fact of life, so trying to forecast their severity is critical.
Every time the earth shakes or a storm lashes the countryside, we are reminded that natural forces are beyond human control. Although science has made big progress in explaining what causes earthquakes and hurricanes, it is still not possible to determine exactly when and where a natural catastrophe will occur and how severe it will be.
For insurers, however, this kind of information is crucial. A catastrophic event usually affects tens of thousands of people and their properties, leading to thousands of claims that may well cost billions of dollars. The 2004 hurricane season in the US, for example, resulted in more than two million claims. According to Swiss Re, more than 300,000 people died last year through man-made or natural catastrophes, and financial losses related to these events reached USD 123 billion. While only 40 percent of losses were insured, this marked the most costly year for the insurance industry to date.
Insurers need to deal with catastrophes well before they occur
Of course, insurers are no clairvoyants. With the help of computer models, however, they can simulate various catastrophe scenarios. The findings play an important role in calculating premiums, determining adequate loss reserves and deciding which risks should be underwritten in which regions.
The fact is, though, that no model can ever be perfect, and however closely weather patterns are monitored, Nature has a habit of springing nasty surprises. The unusual sequence of overlapping storms which hit the US coast in 2004 was a case in point.
But since the introduction of computers, risk modeling has become much quicker and more accurate. In the pre-computer era, modeling a catastrophe was a long and laborious process: nowadays highly sophisticated software programs allow for thousands of different scenarios to be modeled in no time. Of course, before the software can do its mathematics, it needs to be fed with a considerable amount of data about historical disasters and the latest scientific findings. The result is obviously dependent on the quality and quantity of that data.
Hurricane modeling
The course and severity of a hurricane is influenced by countless factors such as water temperature, sea level pressure, wind speeds, geographic extension and topography (further information on hurricanes). With the help of computer simulations, thousands of different storm scenarios with changing parameters can be modeled. Courses of real storms and hurricanes are incorporated with meteorological models. The more storm scenarios are run through, the higher the probability that one of them will match with a real future hurricane. Sadly, 2004 contributed a lot to historical data.
Once the potential course of a hurricane has been modeled, it is projected on to a geographical map that contains data on insured objects. The insurer can now calculate the potential individual loss per object, which is the basis for premium calculations. By adding all individual losses it can get the potential total loss for its whole portfolio. This figure should then be reflected in reserves.
Probability that a catastrophe occurs
Although a model cannot predict exactly when a catastrophe will occur, it can indicate probabilities. Probability not only influences premium levels and reserves but can also be the crucial factor for an insurer to retreat from certain catastrophe-prone regions. In the aftermath of Hurricane Andrew in the US in 1992, 11 local property and casualty insurers became insolvent. Diversification both in terms of geography and lines of business can decide whether an insurer survives a catastrophe.
