Recent trends in mortality and CMI 2019
What you need to know
- At this stage it is impossible to predict the impact COVID-19 will have on future life expectancies, as it will depend both on the number of excess deaths in the UK during 2020 and any secondary effects in succeeding years.
- The Office for National Statistics (ONS) are starting to publish data for 2020 and the impact of the pandemic will emerge over time, but this update focuses on long-term assumption setting.
- The number of deaths in 2019 in England and Wales (E&W) were the lowest there have ever been.
- There was a sharp drop in the number of deaths in 2019 compared to 2018.
- This means that the improvement in longevity in 2019 was the highest it has been since 2011.
- This has led to life expectancies predicted by CMI 2019 to be longer than those predicted by CMI 2018.
- However, the core version of the CMI 2019 model is likely to produce lower life expectancies compared to earlier versions of the model (CMI 2017 and earlier).
- A large variation in both mortality rates and the trends of these over time has been seen between different groups of the population, and this needs to be considered when setting mortality assumptions.
Actions you can take
- Update to the new model for scheme funding valuations and company accounting disclosures.
- Tailor the model to reflect the socio-economic characteristics of your scheme members.
- Consider if the long-term rate of improvement you have previously assumed remains appropriate.
Recent trends in longevity
- Life expectancy has continued to increase since 2011 but at a much slower rate compared to the early 2000s.
- Life expectancy is around 7 years shorter for 65 year olds living in the most deprived areas compared to the least deprived.
- Life expectancy for members of defined benefit (DB) schemes has been increasing faster than that for the general population in recent years.
- Year-on-year mortality is volatile, therefore the particularly low mortality in 2019 could be a ‘one-off’.
The finer detail: How to tailor CMI 2019 for your scheme
The Continuous Mortality Investigation (CMI) published the 2019 version of their model in March 2020. The model is the same as CMI 2018 but allows for death data to the end of 2019.
The CMI model projects life expectancy (also referred to as longevity) by using trends in past E&W population data to predict how it will change in the future.
Future improvements start at the ‘current’ annual changes and are assumed to follow a path to the long-term rate (LTR) which has to be set by the user. The initial addition and smoothing parameters can be amended at the option of the user.
This parameter is the annual rate of improvements in longevity that the model tends to in the long-term and must be set by the user.
The Pensions Regulator’s 2019 Scheme Funding Analysis survey showed that 66% of schemes in deficit that had valuations in the year to September 2017 used the CMI model with a LTR of 1.5%, with 83% using a value of 1.5% or above.
However, the LTR should be specific to your scheme and, as seen in the past, the rate of improvement can vary across different sub-sections of the population.
We can help you tailor this assumption using real-life scenarios to help you understand what may impact on future changes in life expectancy.
This parameter was introduced in CMI 2018 to adjust the current level of improvements to reflect that life expectancy in some groups of the population may be increasing at a higher rate than the E&W population. Subsets of the general population, such as DB pensioners, have been shown to have had higher increases in life expectancy in recent years compared to the average general population.
Therefore it may be necessary to adjust this parameter above the default value of 0% to ensure that your assumption is sufficiently prudent. The CMI are urging users to consider adjusting this parameter to be suitable for purpose.
We can help you tailor this parameter by analysing the socio-economic profile of your scheme members.
This parameter was introduced in CMI 2016 so users can put more, or less, weight on recent mortality experience. This parameter was sometimes used to allow for differences between groups of the population, but the initial addition parameter should now be used to tailor the model for this purpose.
For further information, please get in touch with Natasha Hill or Steve Leake speak to your usual XPS contact.