This could be the first time that trustees and sponsors of defined benefit pensions schemes feel they have sufficient evidence to reflect recent trends in their life expectancy assumptions to some extent. However, care needs to be taken to make sure assumptions are set appropriately taking into account pandemic experience and scheme specifics, which we explore in this second edition of the mini-series.
The typical pension scheme mortality assumption has two elements: the starting level (‘base mortality’), and the path of future change (‘future improvements’).
Depending on the size of your membership and your budget, you might set your base mortality assumption with direct reference to your own membership’s recent experience or more broadly using wider data and taking into account your scheme’s membership characteristics, like pension amount and/or postcode.
Under the first approach, scheme-specific member experience, you need to take care over how to deal with the experience during the pandemic period. Since more members than usual are likely to have died during the pandemic, using this data could lead you to take an overly pessimistic view of the current level of mortality. The solution is typically to exclude the data in 2020 and 2021 from your analysis or otherwise make an adjustment to match how you expect mortality to have increased in those years (making sure any adjustment for the expected COVID-19 impact reflects the characteristics of your membership, for example by age). Data from 2022 can, however, usually be incorporated in a more straightforward fashion if, like the Continuous Mortality Investigation (CMI), you think this is, to some degree, indicative of expected mortality going forwards.
Under the approach using wider data and taking into account pension amount and/or postcode, life tends to be easier since the models used will not yet include data from the pandemic or have been automatically adjusted to allow for it.
Under either approach, make sure you question your actuary over the approach they have taken to get a good understanding of why it is appropriate for your scheme’s situation.
Most, if not all, defined benefit pension schemes will use the CMI’s mortality projection model to predict how mortality rates will change in future. The CMI has allowed for 2022 data when calibrating the 2022 model, albeit with only 25% weight which has led to the headline reduction in life expectancy from age 65 of around half a year[1]. As with base mortality, trustees and sponsors need to decide whether they agree with the CMI’s judgement that the elevated level of mortality in 2022 and 2023 will continue for the short-to-medium term rather than reverting immediately to pre-pandemic levels. The mortality projection model can be tweaked to suit alternative views, for example by adjusting the weight put on 2022 data from 25%, or by changing the long-term improvement rate parameter.
Scheme specifics are also important here. Analysis by the CMI shows that members of pension schemes have been less affected by the pandemic than the general population and that the difference is most notable in the highest socioeconomic groups . Those schemes with better off populations may consider using the initial additional to improvements (or ‘A’) parameter to offset some of the reduction in life expectancy that will flow through from the CMI’s 2022 model.
If anything, the CMI is guilty of making the core projection model too flexible through the number of available parameters and you need to lean on your advisers to ensure it matches your views. There is also an ‘advanced’ projection model available from the CMI but few pension schemes will seek to tailor their mortality assumptions to such a high degree.
Read the next part of this mini-series: Implications of post-pandemic mortality trends for UK pension scheme management.
