The business of estimating risk
Don Jones (L) and Steve Vance install a sheet of plywood over a window of a beach house to protect it from the winds of the approaching Hurricane Earl in Buxton, N.C.
TEXT OF INTERVIEW
Kai Ryssdal: It's going to be a lousy weekend all up and down the Eastern seaboard. Hurricane Earl's a comin'. Surf, wind, rain, and according to one risk forecasting firm, anywhere from $100 million to half a billion in damages. Not by any stretch small change, but also not as high a figure as you might expect for a storm this big. The catastrophe risk modeling firm EQECAT is where that number comes from.
Bill Keogh's a senior vice president there. Welcome to the program.
Bill Keogh: Thank you.
Ryssdal: This $100 million figure that you are forecasting for potential losses from this hurricane, how do you know? Where did that number come from?
Keogh: The simple version is three processes. First, we have to know where the event's going to strike. Second, we want to know what's it going to hit. And third, we want to know what's the insured value of the things that it's gonna hit.
Ryssdal: For the "what's it gonna hit" part, do you guys know, like, the buildings are masonry here and steel construction there and woodframe there? I mean, are your models that precise?
Keogh: Understanding the insured building stock is a very important aspect of building models and something that our clients use. So we use many sources, both public and proprietary, to come up with a view of what the insured properties like in all the hurricane states in the U.S.
Ryssdal: My guess would be that you would prefer not to use words like "accuracy," you know, "we can predict with 77 percent accuracy." You're providing estimates, right, considered estimates?
Keogh: Exactly. What we're trying to do is set rational expectations about the risk, and there's a tremendous amount of uncertainty involved in coming up with these estimates. So, for instance, related there's uncertainty of the track, of the intensity and the wind field of the event. And then for the building, the buildings have different responses. A woodframe building getting hit by peak coast winds of 120 miles per hour might have 30 percent damage or it could have 70 percent damage.
Ryssdal: It's a pretty big window there, 30 to 70 percent.
Keogh: It is.
Ryssdal: And you just sorta have to live with that if you're in the insurance and risk forecasting business?
Keogh: The most important thing we do is we capture and express the uncertainties. It's more about making sure that we're understanding, acknowledging and expressing all the uncertainties involved in the estimates, so that we understand the full range of possibilities.
Ryssdal: So, it's a very big "what if?" question, right?
Keogh: It is. But you know, it's better than nothing...
Ryssdal: Now there's a sales pitch.
Keogh: Well, models are very useful tools. And as long as you understand that it's a tool, you can use it intelligently. If you think it's reality, that's where you get lost.
Ryssdal: Now, as for the companies who might use this data to estimate the risk that they face -- and I'm thinking now of insurance companies -- what do they do with your data once they get it?
Keogh: One of the things they might do is use this information to help deploy claims agents. The sooner you can get to a loss and help mitigate and pay the claims, the better off your client will be. They'll also be using it to come up with a loss estimate for themselves to understand and have expectations for "how large is this going to be for us, how much impact will this be for us?" So they can communicate it to various stakeholders, like shareholders, regulators and insurers.
Ryssdal: Mmmhmm. Now, you're in New Jersey, right? It's conceivably in the path of this storm, yes?
Ryssdal: So how's your insurance, you good?
Keogh: I think we're good.
Ryssdal: Alright. Bill Keogh at the catastrophe risk modeling firm EQECAT. Bill, thanks a lot.
Keogh: Thank you very much Kai.