Mortality projection models are in need of a radical overhaul to ease insurers' funding problems and give them confidence when pricing future retirement income solutions, an influential new report argues.
The study by Swiss Re - A window into the future: Understanding and predicting longevity - suggested current models made insufficient use of available information and could not be easily adapted to reflect medical progress.
It said the key was to use commercial data such as that on the General Practice Research Database to build a picture of how combinations of diseases have affected overall mortality.
It suggested combining this historical information with forward-looking models, considering treatments in trial, to give a series of different events which could be used to describe the reasons for predicted changes in mortality.
This meant when medical advances were realised, it could be seen whether they had already been factored in. If they had not, projections could be updated accordingly.
Report author Daniel Ryan, the head of life and health research and development at Swiss Re, said unforeseen events - such as health scares - could alter projections, but he said this was addressed by calculating a range of long-term assumptions on a disease by-disease basis.
Ryan said this type of approach would have been better able to predict the dramatic improvements in longevity in the 90s.
"We still would have had to change our expectations, but it would have been over a more gradual period of time rather than the kind of step change we saw," he explained.