But chief executive Luzi Hitz says the size of the event is the biggest contributor to loss creep figures

The effects of loss creep could be mitigated by the collation of broader natural event loss data over a longer period of time.

This is according to PERILS chief executive Luzi Hitz, who said more data over a wider geographic area could help risk modelers generate more accurate early loss estimates.

PERILS currently collates industry loss data for 16 countries, with plans to extend this reach by adding New Zealand and Japan.

Reinsurers have been hit by unexpectedly high loss creep figures this year, particularly due to Typhoon Jebi and Hurricane Michael. PERILS does not cover the countries most hit by these catastrophes.

“If we would have done this type of work for the last ten years, also for Japan then eventually yes you would have a better idea of loss creep because then you would have quite a large database of past events, which you can feed into the risk models, which are then used for an early estimate,” Hitz said.

“These risk models then might have been more accurate in their early estimates by having reliable data from the past 10 years.”

However, he said the biggest contributor to loss creep was the size of the event.

“Typhoon Jebi was a massive event that hasn’t happened for a long time – a once in a generation event,” he said. “After that then memory fades and you have new people that have never faced this type of catastrophe, and then there might be a tendency to get it wrong at the beginning. But over time you get it right.”