'EQECAT doesn't do predictions', president Keogh says - it's all about managing uncertainty

Catastrophe modelling firm EQECAT held its London Catastrophe Modelling Conference this week. Global Reinsurance met with EQECAT president Bill Keogh to discuss the latest modelling innovations ahead of US hurricane and what lessons are being learnt from Q1 catastrophes.

What’s new in EQECAT models this year?

We are delivering the ability to model wind and flooding separately through our Asia typhoon model. So (re)insurers with separate portfolios that only cover one aspect can now have a breakdown of risk.

Our US hurricane model is also being refined. We’re introducing a new 16 element time-stepping wind field, which incorporates losses from recent events. We have also introduced the last few years of frequency into our loss assumptions. The net result of this will be losses coming down by 10% or less.

What predictions does EQECAT have that might help reinsurers in the upcoming Atlantic hurricane season?

EQECAT doesn’t do predictions but we certainly are listening to other forecasters’ predictions for this season. But the most useful predictions won’t come out until July.

What we can say with a fair degree of confidence is there should be a higher than normal amount of basin activity. What insurers want to know is if they will make landfall, which can’t be predicted. The 2010 season was a good example of this basin activity. If all that activity had happened just 100 miles further to the west, we would’ve seen probably tens of billions of dollars worth of damage.

Why have the last few years been classed as ‘above average’ seasons?

We are in a period of increased sea-surface temperatures in the Atlantic, which started in 1996. What we have observed is that during periods of increased sea surface temperatures, there is a definite correlation between the increased temperatures and increased frequency and severity of hurricanes.

After recent severe tornado activity in the USA, are there any correlations between hurricanes and tornadoes that might indicate what to expect for hurricane season?

No. It would not be unusual to have a very active tornado season and an inactive hurricane season and vice versa.

Are there any correlations between other catastrophes?

There is research into global plate tectonics but we haven’t got to the stage yet where correlations can be drawn between earthquakes across large distances. We are releasing a new Canadian earthquake model in June that will include those correlation affects between the USA and Canada. But for larger distances, we are still studying and trying to understand the correlations.

What has the scientific and modelling community learned from 2011 disasters?

Scientists studying the Tohoku event have realised that we thought we had such a clear understanding of the risk because GPS systems had been placed over known fault systems. They use the GPS system to understand how much energy is being stored and infer the energy that will be released. But what the scientists have now discovered following 11 March is a lot of the tectonics activity actually happens underwater, which is impossible to observe using GPS technologies

Is the end goal to one day be able to accurately predict catastrophes?

No, not for modellers. Our job is to set rational expectations about risk and it should be used as a framework for understanding and managing uncertainty. The uncertainty simply comes from the phenomena that we are studying and catastrophes like earthquakes will always be uncertain. Predictions are definitive. I suppose from our client’s point of view, it would be nice to know that a hurricane is coming in a week but I am not sure what they would do with that information.