« Climate Change »

Hurricanes, Catastrophe Models, and Global Warming

Yesterday’s Boston Globe had an interesting article about catastrophe models in the insurance industry in the context of uncertainties about hurricanes and global warming. The article raises a number of unanswered questions. Here are a few excerpts and a few of my reactions:

An influential but little known segment of the insurance industry is considering whether climate change might be partly to blame for more intense hurricanes in the North Atlantic. The result of this examination, which comes as scientists debate the same question, could be skyrocketing insurance rates in coastal regions from Maine to Texas.

Already, one leading company that forecasts the risk of natural disasters for the insurance industry has revamped the computer model it uses to simulate future weather trends. The model, which looks five years out, now captures the possibility that global warming might be contributing to hurricane activity.

Risk Management Solutions released the new model in May, predicting that average annual insurance losses will increase 25 to 30 percent in the coastal Northeast because of increased hurricane activity.

Florida state officials are researching whether they should add a climate change component to an insurance hurricane risk model they have developed. And Boston-based AIR Worldwide Corp., another top risk modeler, is launching a study with Massachusetts Institute of Technology scientist Kerry Emanuel to understand global warming's impact on hurricanes and whether insurance risk will rise as a result.

"It behooves us to research this in a scientific way," said Karen Clark, president and chief executive officer of AIR Worldwide. "We want to quantify the effect of global warming on hurricane activity."

The confidential risk models that private companies like AIR and Risk Management Solutions develop are key factors in the price of homeowners insurance bought by many coastal residents. The modelers calculate the risk from hurricanes, earthquakes, and other natural hazards to homes and businesses in a region that takes into account everything from construction material to wind speeds. Insurance providers often use the predictions as an important piece in a complicated formula to set rates.

If global warming is driving more intense hurricanes -- and more of those hurricanes hit land -- it could drastically increase the risk of property loss along crowded coasts.

If catastrophe models are an important factor in insurance rates, and insurance rates are an important factor in insurers bottom line (not to mention a hot-button political issue in U.S. coastal states), then it seems obvious that there is potential for financial conflicts of interest in this area. With respect to pharmaceuticals generally there has been much concern, appropriate in my view, about the role of financial ties to industry among researchers and advisors. And on the climate issue industry funding from the energy sector is tantamount to a scarlet letter. How should we think about insurance industry funding of research related to global warming and insurance risk? [Disclaimer- A few years ago I had a graduate student funded by an insurance company to study uncertainties in catastrophe models.]

Howard Kunreuther, an expert on risk and insurance at the Wharton School, hits the nail on the head when he in quoted in the article:

Ultimately, the problem modelers face is figuring out a short-term prediction from a long-term trend. "The problem is that scientists talk about climate change in terms of 25, 50, or more years; they are not willing to make predictions about five years," said Howard Kunreuther, co-director of the Risk Management and Decision Processes Center at the University of Pennsylvania's Wharton School. "The insurance industry is most interested in knowing what is likely to happen in the next few years as they determine what premiums to set on their coverage against hurricanes and other natural disasters."

Predictions about the long-term future are of course safe, because they cannot be evaluated in the short term. And there will always be this or that event that is "consistent with" the long term predictions, and absolutely nothing is inconsistent with them. According to the article, some scientists are apparently willing in private to make short-term predictions for the insurance industry (also discussed at length here):

In part to deal with this problem, Risk Management Solutions convened a panel of four specialists, including Emanuel, in Bermuda last October to discuss, among other things, what was causing recent hurricane activity and how many storms might hit land.

Aided by the scientists, RMS came to the conclusion that the current period of hurricane activity is so different from the long-term record it didn't make sense anymore to base its models on only the past. In May, the company announced a new model that incorporated the specialists' opinions and the more recent spate of hurricanes, among other changes.

Telling the rest of us what they told the insurance industry, in the form of peer-reviewed, scientific, short-term predictions, would be good in a number of ways. It would allow for empirical evaluation of the predictive skill of short-term (5 years or less) hurricane/climate science, based on actual events. And importantly, it would provide some transparency and accountability for the insurance industry as it ventures into the complicated, conflicted, and political world of climate science, with implications for their bottom line and their customer's insurance rates.