Risk Management Solutions Inc.'s recent upgrade to its U.S. hurricane model incorporates significant new data sets and scientific developments. Business Insurance Associate Editor Jeff Casale spoke with Newark, Calif.-based Ryan Ogaard, senior vp of model product marketing at RMS, about the model and its impact.

Q: What prompted the upgrade of this model and timing of its release?

It's been a three-year process in developing the model, and you get a buildup of data and information that needs to be given to the clients....We were targeting February because there is big renewal season for clients coming up in June and July, and we were trying to get this model out well before then.

We last upgraded the hazard model in 2003 and, as you're doing that, information is still rolling in. It takes years to digest storm information and parse it out and understand what happened in any given storm. We upgraded our vulnerability study in 2006 and we still hadn't included information from 2005 hurricanes because there wasn't enough time to do the analysis. There is an ongoing processing of information at RMS, but we decided that we had enough information available that many aspects of the model would be best rebuilt with this added information in.

Q: What type of data is used in this upgrade and is it different than past hurricane models?

Well, there is certainly a lot more data. There is about 10 times more wind and surge observations from various sources in the study. There is not one single source of information that drives the whole model. Some (information) we buy, some is created by academic or government entities. More information has been gathered about all the recent storms than there has been in the past, so there's been sort of a sea change in how much granular continuous data you can get from a storm. We've also had (hurricanes) happen throughout our gathering process and, with that, the amount of claims and exposure information on individual buildings. We had $18 billion in individual claims data that went into the vulnerability function of the model. Through this, we were able to create a much more nuanced set by region and by type of construction and occupancy class of vulnerability functions than we've had in previous models.

The model also differs in the order of magnitude of more information. There is more information available now than there ever was by an order of magnitude. I guess one more difference is that there have been different expert solicitations that we had, especially in the building code and construction quality area, where we had outside engineers analyze various construction practices and building code enforcement throughout the nation. We wanted to see how effects from a hurricane would or wouldn't occur throughout the nation.

Q: Do you think the data and findings in the model spur changes in the way buildings are built in hurricane-prone areas?

We are one influence among many, and there are a lot of engineering studies in place already that point to better building codes and enforcement. We are just one data point on this and I wouldn’t want to overstretch our influence on that.

Q: Why was it important to understand the way hurricanes deteriorated over land?

We found evidence that certain kinds of hurricanes in certain areas have wind speeds that go further inland than what we had previously represented in our models. There are a number of types of hurricanes that keep wind intensity longer than we previously registered, and there are a number of sources that point to this. The major source was the largest numerical wind field study, which was something that we sponsored and was one of the largest studies of its kind. It basically takes the physics of hurricanes and runs computer simulations of them. The wind fields in the model now are in that study. We feel this is the best actual representation (of hurricanes) we've ever been able to create. One reason for that, there is a lot more data in these models and the computer power is so much more cost-effective now that you can run supercomputers to do (these simulations). In 2003, this wasn't a practical approach to the model, but technology has made it that way now.

Q: What kind of impact does the model have on insurance rates?

Once again, the study is just one influence, but it doesn't trump supply and demand. There are lots of sources for good information for reinsurance rates and insurance rates....The direct connection that a lot of people make between the model and rates really oversimplifies the issue. We're an influence, but we really don't drive the rates per se. I think a lot of people really don't understand that.