Broker Guy Carpenter has published a report looking at effective uses of catastrophe models by property insurers, uncertainty in risk modelling and advocating a multi-model approach to catastrophe models. The report suggests that a multi-model approach is better for estimating risk and controlling uncertainty, particularly given the ever-present factor of uncertainty in risk modelling.
The catastrophe bond market has been moving towards a multi-model view of the world, at least internally. We know of sophisticated investors using multiple models to help them truly understand the risks they are putting their capital to work for. For reinsurers, multiple models are common place as many have their own internal models as well as licenses for the major commercial models. The greater the level of modelling diligence undertaken the better the understanding of the risk and the easier it is to be more certain that risk has been priced correctly.
John Major, Director of Actuarial Research, GC Analytics®, said, “Despite considerable refinement of catastrophe models since their introduction in the late 1980s, uncertainly remains – and it is a significantly bigger factor than many users may realize. While advances over the years have reduced the band of uncertainty around a typical probable maximum loss estimate, the consideration of smaller areas of geography only introduces more uncertainty.”
Guy Carpenter says that uncertainty in risk models is still a larger factor than many model users realise. Back in 1999 Guy Carpenter analysed uncertainty in risk models and concluded “A two standard error interval (a plausible range that has a 68 percent chance of including the true, but unknown, value) for a national writer’s 100-year or higher probable maximum loss (PML) goes from 50 percent to 230 percent of the PML estimate produced by the model.” That uncertainty is slightly lower today, due to technical improvements and better data availability, but the graphic below clearly shows the ‘band of uncertainty’ that still exists in modelling today.
Appreciation of the uncertainty inherent in risk modelling is growing amongst re/insurers and Guy Carpenter point out that as the appreciation of model uncertainty increases re/insurers might find that index-linked or parametric risk transfer products appear more attractive than in the past. They say that basis risk will appear much less significant when put against a background of the uncertainty in the true level of coverage provided by indemnity coverage. That should also make insurance-linked securities and catastrophe bonds more attractive as well and is one of the reasons that parametric triggers have been looked on more favourably.