technology advance to the point that we actually can take what had only been a capability reserved for mainframes and looking at risks to nuclear power plants and things like that and we’ve been able to evolve that into something that can be used on a desktop and can be used to make decisions nearly instantaneously now and inform many kinds of risk decisions.
The models are dependent upon the science, whatever they’re trying to model. All three of us talked about some of the challenges in modeling these three perils. They’re also dependent on the quality of information that’s being put into them. Depending upon the peril, we have more or less information about what it is that’s trying to be modeled, and how it will perform in those events. We’ve learned a lot since 2004 in particular about how buildings are performing to a number of different wind speeds and in different kinds of climatic conditions. We still haven’t had a lot of testing on the earthquake side and even less so on the terrorism side.
There are also still a lot of challenges in data quality. What drives a model is knowing the location, knowing the kind of structure or asset that’s being evaluated, and with that goes the proper policy terms and limitations across all those lines. You can reduce your uncertainty tremendously by improving your data quality. It’s tough to bash just one piece of the modeling process because, really, that the user is just as involved in that output as the model.
McDONALD: Brian, there are a number of terrorism models. The purpose of terrorism is to do the unexpected. What do you see in terms of how models have been developed and particularly how they’re being used?
FINLAY: Obviously this is an incredibly difficult space to talk about the use of models for the reasons we discussed earlier. Laurie just recited it far more eloquently than I could. That is it at its root. We’re talking about models that rely upon, obviously the decisions of individual beings. They are not rational actors that are perpetrating these incidents.
This past year a congressionally mandated commission released its report in Washington. It’s the WMD Commission on Prevention of WMD Terrorism. It took the town by storm to some extent in that they predicted that within the next five years there is a greater-than-even chance of a WMD incident occurring. Not necessarily here in the United States but possibly in the United States. The most likely event in their minds was a biological incident because of the ease of access of biotechnologies and advanced and sophisticated weaponry in that area to terrorists.
On the other hand you have experts, a group from Harvard, that came out and said we think that the probability is right but it’s 10 years, not five years. Others have said: hide under your desks now because it’s coming at any moment. It’s just such a difficult and convoluted space. In some ways we can become numbed to the constant, the sky is falling, the sky is falling. Almost a decade after 9-11, the sky hasn’t fallen.
Any macro level analysis of terrorism suggests that there will be in the offing, whether it’s two years, five years or 10 years, an incident in which– this is just the simple law of probability – a terrorist acquires some sort of weapon of mass destruction. There is little question in anyone’s mind that if and when they do, the target of that attack will be the United States.
McDONALD: Another question came in, a hurricane-specific question. Roger, you gave your views that there wasn’t any increase in intensity or landfall frequency over the past 100 years. This person points to Ike. Not that they’re arguing with you, but they’re noting that it was not very powerful until it came to land. Then the size of it caused it to have a significant impact of course. Has there been any increase in the size or hurricanes in that time frame and also any changes in the paths, the trending of those paths?
PIELKE: What the scientists who study hurricanes have observed are what they call multi-decadal changes in hurricanes. The discussion is that there’s an active period at the beginning of the century, a quiet period, an active period, a quiet period and now we’re in an active period again. Some scientists look at that and say we see a cycle that has to do with the sea surface temperatures in the ocean. Other scientists say we see other signals there, one of them being the effects of the Clean Air Act, which surprisingly changed the aerosols over the ocean, changed the ocean temperatures. There’s some uncertainty as to whether there are these cycles in fact in hurricanes, and if there is, what causes them.
We’ve been in an active period at least from 1995 through 2005. It’s quieter now. My personal view is I don’t think anyone knows if we’re at the start of another quiet period or if it’s just a lull in this more-active period. The inflection point on when things change is really something that scientists are studying but no one has a handle on.
On the other part about the size of hurricanes, if you look at landfall there’s a lot of debate about what happens over the open ocean. If you look at landfall, I haven’t seen any evidence that storms have gotten progressively larger or stronger or any changes in landfall.
McDONALD: Sanjay, here is a more insurance-specific question that plays nicely with the discussion of the E&S market as well. Does it make more sense to have a separate wind market from an admitted carrier that excludes wind? Or does that create potential E&O issues? If you separate it out is there any impact there?
GODHWANI: On the wind side is it’s always better to buy your insurance from one carrier, specifically a significant primary layer because you’re negotiating and managing your claim through one adjuster. So that’s the driving force in the decision. You have one person coming in, or one company, to manage the process of the claim adjustment. Otherwise the policies are relatively separate on what is covered in a named storm or wind event in the sense that flood is covered and a tornado as an act of coming out of a named storm. So that has been buttoned up from a policy language standpoint.
McDONALD: We also received a question that is earthquake-specific and about models. We now have a case study here in front of us: How do models stack up when compared to actual events when they occur, for example, Haiti? What actually happened versus what we thought might happen? Any thoughts there? Is it too early? What are you seeing?
JOHNSON: In Haiti?
McDONALD: Yes.

Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9