Forget frequency, location is everything when mulling wind risk

Re/insurers should not focus too much on event frequency when it comes to windstorm risk—just like real estate, location (where a hurricane makes landfall) is everything, Karen Clark, chief executive officer of Karen Clark & Company (KCC), told Monte Carlo Today.

“We need only one storm to make landfall in a specific place, perhaps downtown Miami, and you are looking at the industry’s first $200 billion loss,” she said. “Frequency matters a lot less than where they land, location is everything.”

Clark was speaking in the context of a key anniversary: the industry recently reflected on 30 years since Hurricane Andrew, an event widely regarded as representing a turning point in the way the industry viewed risk—and its willingness to embrace catastrophe models.

The hurricane in 1992 resulted in the largest insured loss in US history at that time: some $15 billion. It starkly illustrated how poor the risk transfer industry’s understanding was of its own exposures. The industry was taken by surprise as the losses escalated, despite early catastrophe models, which had been developed by Clark, offering relatively accurate estimates early on.

Some five years earlier in 1987, Clark, widely regarded as one of the key pioneers of catastrophe modelling, had founded Applied Insurance Research—later AIR Worldwide. She notes that it is a common misconception that cat models were developed in response to Hurricane Andrew. She does agree, however, that the event made reinsurers in particular sit up and take notice of their potential.

“Even in 1992, some 20 reinsurers were using our cat model, but even our own clients didn’t initially believe the numbers generated by the model. It turned out that its initial estimates were almost spot on,” Clark said.

“It illustrated another stark reality: if it had hit downtown Miami, the loss would have been closer to $60 billion. That sequence of events catapulted risk modelling into the mainstream. Within years, rating agencies were embracing it and the very premise of cat bonds was based on it.”

It triggered a fundamental change in the way the industry sought to collect, track and analyse their exposures. Reinsurers suddenly wanted more data from their insurers. This, combined with ever-more powerful computing, has propelled risk modelling forward to the point that estimates are ever-more granular and available more quickly.

Despite this, Clark believes many reinsurers have become too reliant on a limited number of models. They have invested so much time and money integrating them into their own systems and underwriting frameworks, it has become almost impossible to backtrack. Yet they have also become too inefficient and unwieldy and are not able to be updated quickly enough, given the rapid changes being seen in cat risk due to climate change.

Partly because of this dynamic in the reinsurance world, KCC, which offers its own models, is working more with primary insurers, unencumbered by this historic reliance on a small number of models.

“Our technology is easier to implement and much more responsive, offering automated real-time updates,” she said. “Primary insurers are seeing real value in integrating KCC technology into their business models.

“They tell us that we are more accurate—and they say that based on their actual experience of comparing our estimates with their eventual losses. We update our models more frequently to take into account variables such as climate change and social inflation. They tell us ours is more transparent and easier to integrate.”

She admits that no model is perfect. “It has been said that all models are wrong, but some are useful. I say models that are less wrong are more useful. I believe it is important to embrace new technology and ensure that models accurately reflect the true risk.”