Nicholas Lax, member of the CMI Assurances Committee, explores how the richer, more flexible CMI Model is helping practitioners.
Mortality trend modelling for term assurances is entering a markedly different landscape.
The pandemic disrupted improvement patterns at working ages, widened socioeconomic differentials, and increased the need for explicit practitioner judgement.
Meanwhile, CMI_2024 introduced major enhancements, most notably the fitted overlay in the new APCOI framework, giving users new levers to reflect the COVID-19 spike and runoff in a more age-sensitive way.
Working Paper 212 sets out practical guidance for using this richer, more flexible model to manage basis risk and make post-2019 experience more informative for assured lives.
The CMI_2024 version of the model introduced the most significant methodological upgrades in some time. Chief among these is the fitted overlay component in the new age-period-cohort-overlay improvement (APCOI) model, designed to reflect the transient effects of the COVID19 pandemic.
Earlier versions simply put reduced weight on data from 2020 to 2023, and smoothed through the pandemic years. The new overlay allows users to capture both the spike in mortality and its gradual runoff.
Importantly, users can tailor this overlay by age. This is critical for term assurance portfolios with younger age profiles, for which the peak pandemic impact occurred later than for older insured populations.
There are also other changes to the model, which allow mortality improvements to vary more meaningfully by age. The most significant of these is the introduction of multiple period terms, which allow for a better reflection of changes in mortality improvements at different ages.
This is particularly important for term assurance users, because recent trends (both pre- and post-pandemic) at younger ages differ noticeably from those at older ages.
The model changes, such as those described above, give practitioners more levers, and a greater reliance on their informed judgement to pull the right ones. The new structure provides users with the ability to make choices about:
Importantly, these flexibilities allow practitioners to better reflect age-specific trends in their modelling.
A key theme in our guidance is basis risk. The model is calibrated to the general population, but term assurance lives differ in many systematic ways:
All of these factors mean that population-level improvements will rarely show the full story for term assurances. Adjustments, particularly to initial improvements, may be necessary to reflect the expected trend for assured lives.
The new model allows for the more effective use of post 2019 experience in setting assumptions, but with care. The new fitted overlay means users can now include pandemic mortality in expected claims calculations rather than ignoring this data. This provides a clearer, more consistent way for practitioners to assess experience.
For practitioners, the overarching message is that the model is now richer, more flexible, and better suited to capturing the complexity of mortality improvements for term assurances. But using it well requires clear judgement, transparent assumptions, and an understanding of the unique characteristics of term assurance lives.
The Mortality Projections Committee plans further work in the first half of 2026 on how base mortality and improvement assumptions interact. This will include examples of pitfalls and offer potential solutions.
