*
*
OPERATING AND FINANCIAL REVIEW / THE COMPANY / DELIVERING VALUE /

MODELLING

*
*

We use modelling to achieve four aims:
  • To help price individual risks;
  • To manage and balance our portfolio of risks;
  • To help ensure that our risk remains inside the Group’s risk appetite;
  • To help quantify our loss after an event.

2004 will be remembered for an almost unprecedented Atlantic hurricane season with four major storms making landfall in the United States and affecting Florida. The total insured mainland US property damage is estimated at US$22.5 billion. Risk modelling agencies variously calculate the likelihood of four hurricanes making landfall in Florida as once in every 204 to 325 years.

In the immediate aftermath of each hurricane Amlin’s risk modelling team was rapidly able to estimate the possible loss cost to Amlin. Preliminary and latest estimates of Amlin’s reinsurance losses from each of the hurricanes are as follows:

PROPERTY INSURED LOSSES $m

Event Prelim. gross
loss (8.10.04)
Latest
gross loss
Prelim. net
loss (8.10.04)
Latest
net loss
Charley 61.2 65.7 22.3 27.4
Frances 63.0 90.4 34.0 34.7
Ivan 70.1 79.2 31.9 34.9
Jeanne 51.4 29.7 26.8 21.6

With Hurricane Jeanne taking a similar path to Hurricane Frances there has been difficulty in assessing which of the two caused some of the damage and generally, where there has been doubt, policy holders have attributed their losses to Hurricane Frances. Also, loss costs have generally been higher than originally predicted because, with the overall quantity of damage and resultant shortage of contractors and building supplies, expected rebuilding costs have risen.

Amlin has a dedicated catastrophe modelling team. The group was expanded during 2004 and acts as a key link with both the underwriting and claims teams.

We use a number of proprietary models including AIR CATRADER® and RMS RiskLink®. The systems combine the capture of exposed aggregates with catalogues of thousands of scenarios built from both historic and simulated events. These “catalogues” contain not only hurricane events, but also earthquakes, severe winter storms, tornados and floods.

We continually model a range of major exposures, by cause, location and size or probability. Some examples, with changes in possible loss impact over the last three years, are shown below:

Events Current
modelled
probability
Net loss to Amlin as a
% of net tangible assets
(at 31 Dec)
(years) 2002 2003 2004
Major Los Angeles earthquake 1 in 250 23.7% 23.8% 18.6%
Major US windstorm 1 in 250 30.5% 21.3% 15.3%
New Madrid earthquake 1 in 500 30.5% 21.3% 15.3%
European storm 1 in 100 33.9% 38.3% 27.3%
Japan earthquake 1 in 250 29.9% 23.0% 23.1%
Major aviation collision n/a 22.1% 17.3% 15.0%
Major marine collision n/a 16.0% 18.5% 12.8%
Major UK terrorist event n/a 26.3% 15.3% 11.7%

Catastrophe modelling is not an exact science and the agencies are constantly looking at new data to improve modelling accuracy. The above therefore can only be used as a guide.

Back to previous page


COPYRIGHT AMLIN2005



Back to top  Print