Pension schemes and insurance companies use longevity (life expectancy) projections to place a value on their liabilities, and insurance companies also use them to set annuity prices. CMI has produced a mortality projection model on an annual basis since 2009, reflecting newly-published mortality data by the Office for National Statistics (ONS). The next version of the model will be published in early 2017, and CMI has issued a consultation on proposed changes in the model structure.
This review takes place following a few years where the number of deaths in England & Wales was above previous projections, with the average mortality improvement over 2011-2015 being the lowest since at least 1975. In particular, ONS data shows that 2015 was an exceptionally heavy year for mortality, with over 36,000 more deaths than the expected 500,000 in England and Wales. 2016 has been less extreme so far, but mortality is still above trend, with around 14,000 more deaths in the first 23 weeks of the year than the expected 230,000. This means that the life expectancy projected by the latest version of the model is lower (by around a year) than models produced in earlier years. This means that, effectively, there have been no significant improvements in mortality rates since 2011.
When the previous version of the model was released in September 2015, Aon Hewitt cautioned against relying on it too much, noting that it believed the model might be too responsive to short-term impacts on mortality rates. With its new proposals, CMI has made the model less responsive to single years of data (which may be caused by one-off factors), while still responding appropriately to long-term changes in mortality expectation. CMI also proposes to give an option to practitioners to 'dial up' or 'dial down' the responsiveness of the model to new data.
As part of their review, CMI has also aimed to make the model more transparent and easier for practitioners to calibrate and use.
Martin Lowes, partner at Aon Hewitt, said: “Longevity projections are key for future planning for both pension schemes and insurance companies. We were previously concerned whether the CMI 2015 Model was over-reacting to recent years with relatively high numbers of deaths, as these may be caused by one-off factors rather than being part of a long term trend. Insurers and pension scheme trustees were certainly concerned about the volatility of longevity projections. For example, moving from CMI 2014 to CMI 2015 reduced liabilities by around 1%. Was that really justified by one more year’s data?
“We therefore welcome the CMI's focus on the responsiveness of the model, as well as its aim of making the model much easier for practitioners to use. The proposals reflect an evolution of the modelling process rather than a revolutionary change. Based on the consultation paper, we believe this model is likely to be a good step forward for mortality modelling.”
|
