Eye of the Storm

DPO “is really our secret sauce,” Karon says. “Reinsurance has typically been a back-end solution,” helping insurers “lay off” risk they’re already taken on. “This is a front-end solution, helping them pick which risks go in the front door in the first place.”

Some of the people in Benfield’s offices have been up to their necks in building these sorts of applications since the world’s first cat model was developed in the 1980s.

The acknowledged pioneer of modern cat modeling is Karen Clark, founder of Boston-based Applied Insurance Research. In 1987, AIR developed the world’s first comprehensive software program to predict insurance losses from hurricanes and tornadoes. Clark, who now heads Boston consulting firm Karen Clark & Company, says that E. W. Blanch was not only her first client but a joint-venture partner in developing an early cat model called Catalyst.

Blanch “was always very innovative,” Clark says. “They realized around the same time I did that the insurance industry needed a better way to quantify its exposure to hurricanes.” As she recalls, someone at Blanch saw an early paper on cat modeling that she wrote for an actuarial society, and called her out of the blue. Before AIR had an actual product, it had a client and development partner.

Today, cat models are standard tools, licensed by reinsurance brokers, insurance companies, and reinsurance companies to quantify risk. AIR is now one of three major model suppliers; the others are Risk Management Solutions, Inc., and EQECAT, both based in California. All three models assess risk from hurricanes, earthquakes, and other disasters, says Wendy Hayes, Benfield’s senior vice president of catastrophe modeling.

Hayes and her group feed a client’s policy data into these models, then run scenarios that show the effects of thousands of possible “events”—hurricanes of different strengths, for instance, hitting the U.S. mainland anywhere from Texas to Maine. Each event is assigned a certain probability of occurring; a so-called 100-year hurricane, for instance, is actually a storm with a 1 percent probability of striking in any given year.

Based on that analysis and on more detailed modeling from applications like Exposure View and DPO, Benfield advises clients about how much risk to lay off by selling it to third parties. Karon says the structure of those deals depends on a lot of things: how much capital the insurer has (“and frankly, whether they’re publicly traded or not, because many publicly traded companies need to have more stability”); their need to maintain good ratings from agencies like A. M. Best and Standard & Poor’s; management’s “risk appetite”; and finally, market conditions—how much appetite the buyers of risk have, and how much reinsurance costs as a result.