The Continuous Mortality Investigation (CMI) Mortality Projections Committee (MPC) publishes annual updates to the CMI Mortality Projections Model. This blog summarises the latest version, CMI_2022. There is more detail in the:
The CMI Model is used by UK pension schemes and insurance companies which need to make assumptions about future mortality rates. The CMI Model projects mortality improvements (i.e. annual reductions in mortality rates) by interpolating between current mortality improvements and assumed long-term improvement rates. The current mortality improvements are calibrated to historical data for the general population, but users are encouraged to consider adjusting the Model’s parameters so that it reflects the specific population that they are using it for. Users must also form their own view of long-term improvements.
Standardised mortality rates in England & Wales in 2022 were on average 3% lower than in 2021. However, mortality in 2020 to 2022 was significantly higher than before the coronavirus pandemic. Compared to mortality in 2019, mortality in 2020 was 14% higher, mortality in 2021 was 9% higher, and mortality in 2022 was 6% higher.
While mortality experience in 2020 and 2021 will affect actuarial calculations, mortality in both of those years was exceptional and is unlikely to be indicative of future mortality. For this reason, the CMI places no weight on the data for 2020 and 2021 in the core version of the model. While mortality in 2022 has also been higher than pre-pandemic levels, it has been less volatile and may be indicative of future mortality to some extent. Given that, and following a consultation with users of the model, the CMI has placed 25% weight on data for 2022 when calibrating the core version of CMI_2022.
The data used to calibrate CMI_2022 has been revised to reflect results of the 2021 census. We have used the ONS population estimates for 2021, which take account of the census. Unfortunately, the ONS has not yet published revised estimates for 2012-2020 – these are now expected in September. We have therefore made our own revisions for 2012-2020, the period between the two most recent censuses, to be consistent with the published 2021 figures. Given data limitations, we have opted for a simple and pragmatic approach for our revisions. We intend to use the revised ONS figures in future versions of the Model when these become available.
CMI_2022 produces cohort life expectancies that are considerably lower than in CMI_2021 – cohort life expectancies at age 65 are around 7 months lower for males and around 6 months lower for females than in the previous version of the model. This fall is mainly the result of the relatively high mortality in 2022, but the revised population estimates for 2012-2021 account for around 1 month of the fall in life expectancies at age 65.
Cohort life expectancies at age 65 for CMI_2022 are around two years lower than in the first version of the CMI Model, CMI_2009, for both males and females, reflecting the slowing of mortality improvements since 2011 as well as the impact of the pandemic.
We will continue to monitor emerging mortality and review the Model so that CMI_2023 meets Subscribers’ needs. We expect to increase the weight placed on data for later years and gradually return to 100% weight over a number of years, but we have not made a firm decision on this yet. We aim to communicate our plans for CMI_2023 in the next “interim update” working paper, towards the end of the year.
We delayed the release of CMI_2022 to June 2023 with the aim of allowing for revisions to population estimates which were expected to be published by the ONS earlier in 2023. We expect to revert to the more usual March publication schedule and publish the next version of the Model, CMI_2023, in March 2024, subject to the continued publication of provisional weekly deaths data by the ONS and timely publication of the revised ONS population estimates for 2012-2020.