By MATTHEW BRODSKY is senior editor/Web editor of Risk & Insurance®.
Before I begin, let me just say that I have plenty of friends in the catastrophe modeling business. I might also add that I am not going out on a limb here in attacking their niche industry. It is way too easy anymore to criticize catastrophe models.
We all know that the software vendors that hawk this insurance technology are ghouls that live for disasters past, present and future, putting out press releases about insured loss estimates before the earth stops rumbling and the storm surge stops flooding.
And we all know that their loss estimates are always inaccurate and need to be updated. Their products are based on uncertain science, represent a limited view of reality and are generally populated with spotty data. What do you expect?
The catastrophe modelers will be the first people to warn you about this "uncertainty." The other word they like to use is "assumptions," as in, the three models (from RMS, AIR Worldwide and Eqecat) all produce different loss estimates because of different assumptions. Yet the catastrophe modelers still force their wares on us, all the while never making entirely clear to anyone what in particular these uncertainties and assumptions are.
Where we stand now, insurers and reinsurers must use these models to estimate their exposure to catastrophes, to determine how to price risk and to calculate how much to set aside in reserves. Not only do all their competitors do the same, but ratings agencies demand it.
The term "black box" is cliché, so let's use a different analogy for the models: that of the cymothoa exigua. Also known as the tongue-eating louse, this crustacean creeps into a spotted rose snapper's mouth and purposefully causes the poor fish's tongue to decay. The critter then attaches itself to the nub and effectively becomes the fish's tongue. The dim-witted snapper then goes about its life, seemingly still able to eat its normal diet and thrive, while the parasite gets to live off the fish's blood and mucous. Even trade, right?
The difference with catastrophe models, however, is that there's always a latest version to swallow. Herein lies the problem with RMS 11.0 (and similar, less conspicuous hurricane model updates from AIR and Eqecat). The updates require a major rethink by insurers on how to handle hurricane risk, leading them to move many millions into reserves and reconsider how much their capital is worth.
The losers here, ultimately, will be insurance buyers, who get the costs of all this passed down to them. You see, the fish protests the new models, but ultimately still gets to eat. The tongue-eating louse gets the leftovers it needs to succeed. And the insurance buyers, well, you're the fish food.
(Read Managing Editor Cyril Tuohy's Point, "Embrace Your CAT Models," here.)
August 1, 2011
Copyright 2011© LRP Publications