...the article explains it all. Whislt we may never know the exact causes (the Cat season seemed to have a major effect on the .25c per share loss), it would seem that for some of the reinsurers, the over-confidence in Cat models is coming home to roost.
This is similar to the problems that the "high finance" industry faced after the Russian bond default and the asian crisis of the late 90s. An influx of "uber-intelligent" (read: mathematicians) people swooped in upon the finance sector and started modeling everything in sight based upon extremely sophisticated mathematical models. Most of these models had analogues in physics. Sophistication and complexity somehow lead these "uber-intelligent" people to disregard the underlying economics of the risks involved. In the Russian bond case, LTCM ("Long Term Capital Management") lost billions by not forseeing the economic and political consequences of a sovereign bond default. Maybe they had the financial derivatives priced correctly...but priced correctly for what? A good thematic account of the LTCM:
Now, there is nothing wrong with attempting to calculate probability distributions per se (such is the nature of our lot in life in the insurance business), but there is something very wrong when one's overconfidence leads one to believe that we have the only objective view of the "true" probability distribution.
The central problem is best described by this quote from the great Nassim Nicholas Taleb (for the statistical neophyte "Gaussian"="normal distribution")
"Circulatity of Statistics (The Statistical Regress Argument): We need data to discover a probability distribution. How do we know if we have enough data? From the probability distribution. If it is a Gaussian, then a few points will suffice. How do you know it is a Gaussian? from the data. So we need the data to tell us what is the probability distribution, and a probability distribution to tell us how much data we need. This causes a severe regress argument."
(RE)Insurance could learn alot from that. The central limit theorem cannot be applied to everything.
We should see more of these types of things until the zeitgeist of overconfidence in quantitative methods slowly begins to commingle with traditional risk and financial analysis. We are already seeing the effects in finance and insurance research.