Towards the predictable pandemic

Eric Sugier, head of property at Liberty Mutual Re on how Covid-19 is a proving ground for pandemic models…

2020 has been the very epitome of unpredictability.

Was there anyone in the world who envisaged shopping in facemasks, maintaining social distancing, worrying about something called the R-rate and planning how to re-engineer office space for a new type of working?

And yet, a mere 10 months since the first cases of Covid-19 were identified, here we are living in a very altered world.

It’s fair to say that the advent of Covid-19 and its rapid global spread took all manner of experts by surprise. Not that pandemics are unprecedented, you understand. Over a century ago Liberty was supporting insureds and employees through the Spanish Flu pandemic of 1918. More recently in 2006, RMS launched the market’s first model for assessing the impact of an influenza pandemic following concern over the H5N1 strain.

That model was updated in May to take account of the realities of Covid-19. However, it is the events that are part-modelled, or even unmodelled, which have the power to affect both reinsurance and retrocession markets in ways that are hard to predict and their ability to move the market is much greater.

This summer, reinsurance rates have reacted decisively to the pandemic, mirroring what has been happening in the primary market. Double-digit price increases for property and casualty risks have been reported. Exclusions and restrictions are also appearing in contracts. One commentator noted that the reinsurance market had needed a ‘little push’ to harden: Covid-19 was that push.

In common with market-moving events like 9/11 and Hurricane Katrina, Covid-19 has many elements that have been modelled but others that have not. Take 9/11, for instance, where the scenario of a plane crashing into a building was a familiar one, but not a plane being deliberately flown into a building.

Similarly with Katrina: hurricanes are part of any reinsurer’s staple diet of such power or one that resulted in the failure of flood defences.

“We need to acknowledge that models are but a single weapon in our armoury.”

The models we use to stress test our portfolios and reserves have improved significantly in recent years. They are good tools with which to assess the potential loss contribution of one account to a portfolio. They illuminate an event’s impact on an account and allow us to examine relative valuations from year to year.

But reinsurers must not rely on them too much for an absolute value. If one examines a similar scenario with different model providers, the variations in outcomes expose the differing assumptions the modellers have made in their creations. Even with the same model, there are significant differences from one version to the other. I come from a mathematical background, which should make me a strong advocate for models – but even I take models with a small pinch of salt.

2020 has absolutely highlighted that no one can be prepared for every eventuality – life will always surprise us.

Covid-19 is a proving ground for pandemic models. For the first time in 100 years, our industry and its modellers are experiencing the reality of pandemic and its impact on society and the economy. But we do need to acknowledge that models are but a single weapon in our armoury, and, in doing so, we take an important step in terms of real preparedness.