July 22, 2011 by Canadian Underwriter
Long-range weather event forecasting – from one to 10 years out – would be useful to the insurance industry, which now relies too heavily on historical data to predict the likelihood of severe weather events, according to a report co-published by Lloyds of London and the Met Office (the United Kingdom’s national weather service).
“Our climate is changing, and we appear to be seeing more frequent extreme weather events, so the wisdom of relying solely on historical observations is being debated,” says the report Forecasting Risk: The Value of Long-Range Forecasting for the Insurance Industry. “In addition, the recent past does not capture the impact of rare catastrophic events, such as from large volcanic eruptions, and this is an area where the forecasting sciences may help.”
The Lloyds/Met report observes that recent research has shown that current ocean conditions can be used to inform the likely risk levels of future extreme loss-making events over one to seven years ahead.
The key is in the development of dynamic forecasts.
Dynamic models take numeric forecasts from a computer model of the earth’s weather and climate system – a climate system that takes into account initial atmospheric and ocean conditions – and evolves them into the future following the laws of dynamics and thermodynamics.
“One relevant question is: how good will the forecasts be in five to 10 years’ time?” the report asks. “This is not easy to answer and is dependent on increased scientific understanding of fundamental processes in the climate system, supercomputing resource and the development of [insurer]-relevant impact models and tolls to aid decision-making.”
The report concludes development of long-range forecasting technology is “probably 20 years behind weather forecasting.” But its development would allow insurers and reinsurers to better understand the links between different peril classes once different hazards in different locations are linked by large-scale climactic drivers.
“To a large degree, the industry does not currently integrate long-range forecasts from statistical or dynamical forecast models into core risk pricing decisions,” the report concludes. “Should [long-range forecasts] now be proven to be skilful and provide more relevant information (for example, the likelihood of landfalling tropical storms), in a way that better meets end-users’ needs, long-range forecasts should be useful to the insurance industry for managing risk exposure and ensuring profitable business.”