Sunday, August 06, 2006

Hurricanes: Forecasting and insurance

A recent article in the Sunday New York Times, “Researchers Lower Estimate of Hurricanes This Season”, provides perspective on how little credence the media, politicians, and environmental activists deserve when they confidently forecast either doom or the absence thereof.

Forecasts of the number and intensity of hurricanes on a one-year basis are not accurate and cannot expected to be accurate. Neither are forecasts for several years very accurate.

Hurricanes, in a particular part of the country, such as Florida, are not insurable, i.e., the total loss cannot be reliably predicted for one year or several years.

It may be that hurricanes are not even insurable in the aggregate, at least for time intervals of several years. This is because they are too few in number to average out over only a few years.

The worst hurricanes may not be insurable over any reasonable time interval. By definition, such hurricanes are the worst storms. That means they are low probability events. Trying to average out events drawn from the extreme right tail of a probability distribution takes a lot of draws and the volatility is enormous.

One implication is that hurricane insurance losses cannot be forecast very accurately and insurance companies that charge what the media, politicians, and consumers view as fair hurricane insurance premiums are likely to money.

The same perspective applies to predicting and insuring against most low probability events.

Reinsurance works only if you don’t go too far into the right tail of the probability distribution.

Here are some excerpts from the New York Times article.

Hurricane researchers at Colorado State University predicted Thursday that this year’s hurricane season would not be as bad as their earlier forecast had indicated, and they said a monster storm like Hurricane Katrina was unlikely.

In December, the researchers predicted nine hurricanes, five of them intense. On Thursday, they reduced those numbers to seven and three.

“Moreover,” they said, “these forecasts do not specifically predict where within the Atlantic Basin these storms will strike.”

No comments: