Insights on the 2024 Canadian Pensioner Mortality Research Project

Highlights of the CPM2024 Research Project

Data and effective date

• Significant growth in dataset compared to CPM2014
  The dataset used covers 33 plans, 1.6 million records, 11.3 million exposure years, and over 300,000 deaths across 2011-2021. This is more than triple the volume of data used for CPM2014.
• Effective date of January 1, 2024
  The tables have an effective date of 2024. The study is actually built on pre-pandemic and pandemic data (over the period 2011-2021). However, to aid practitioners, the resulting mortality rates have been projected beyond the study period using underlying mortality improvements observed in the data, after accounting for the impact of the Spring 2020 COVID-19 spike.

Methodology

• Methodological shift to survival models
  The CPM2024 tables are graduated using a modern parametric spline‑based survival model which uses a consistent framework for modelling subgroups. The tables are created by graduating the force of mortality by fitting four key parameters (including time-varying components) to each dataset which is then converted to a probability of death.
• Age extension
  The model extends the mortality tables to higher and lower ages beyond the core data range within the same framework. Older ages come directly from the fitted spline functions, while pre-retirement ages are added later using a smooth transition.
• Seasonality and COVID-19 effects
  The model incorporates seasonal variation in mortality experience, accounts for periods of COVID‑19 excess mortality, and factors in a time trend to reflect the underlying mortality improvement over the study period. The resulting tables should therefore not be biased by the exact time period covered by different contributing sets of data and should not reflect significant spikes of mortality due to COVID-19.

Segmentation of tables

• Addition of survivor tables
  The CPM2024 release includes tables for surviving spouses, recognizing the elevated mortality typically observed among widows and widowers of deceased retirees.
• New segmentation: Heavy, Light, Combined 
  The proposed CPM2024 tables have three categories: Heavy (higher mortality), Light (lower mortality), and Combined (full data). The Heavy and Light categories were segmented by splitting out experience from individual plans with particularly heavy or light mortality experience (not by using specific risk drivers). This segmentation is intended to offer practitioners greater flexibility in selecting a baseline assumption aligned with their membership’s characteristics. As a result, actuaries will need to apply professional judgment and consider plan-specific  mortality characteristics (such as postal codes or affluence measures) or plan-specific experience, when determining which tables most closely match their plan.

• Other risk factors
  The public-private sector split from CPM2014 has not been used in CPM2024. The researchers observe that there is a large amount of heterogeneity among both public and private sector pension plans, so did not see a justification for modeling these plans separately. We note this is consistent with previous Club Vita research on this subject.

  A number of other risk factors known to be correlated with differences in mortality rates, such as pension amount, salary, postal code, occupation, pension form and retirement health were not used in segmenting tables in CPM2024 (due either to availability of data or concerns over uniformity of data descriptions across different plans).

• Un-weighted and weighted tables
  The research only takes into account the impact of pension amount by graduating rates weighted by pension amount. As a result, the release includes two sets of tables: un-weighted (by lives) and weighted (by pension). 

Published tables

The release includes 10 tables in total:

• 4 gender/pensioner type tables (Combined data: male/female retirees, male/female surviving spouses), and
• 6 segment tables (Heavy/Light for all except male surviving spouses)

What should pension actuaries be considering?

Over the coming weeks, Club Vita will publish three deep‑dive articles exploring considerations for pension actuaries and industry practitioners.

1. Further thoughts on CPM2024 tables: How are longevity risk factors reflected?
   The CPM2024 model captures some key risk factors - including separate treatment for retirees and surviving spouses, and the new Heavy/Light segmentation. This represents a significant improvement to the options available in CPM2014, however, these categories reflect only part of the variation observed in Canadian pensioner mortality. We will explore the underlying risk factors that drive longevity differences, such as socioeconomic factors, postal code, occupation, and pension amount and discuss different options available for pension plans that wish to capture the impact of more of these factors than available by solely using CPM2024.
2. Further thoughts on CPM2024 tables: How is COVID-19 reflected?
   The study adopts a thoughtful approach to pandemic experience, by capturing the first COVID-19 peak in the model and fitting to ensure the excess mortality is excluded from the graduated baseline rates. In this article, we will examine the different approaches actuaries may take in making adjustments for pandemic experience: whether to treat it as a temporary shock, adjust for lingering effects, or exclude it entirely from assumption setting. We will compare the study’s approach to Club Vita’s, as well as how the approach will impact implementation.
3. Further thoughts on CPM2024 tables: How do they interact with improvement scales?
   The CPM2024 model includes a time trend component that projects mortality rates to the effective date of 2024. Practitioners need to consider how to pair the new base tables with the current Canadian improvement scales for the purpose of experience studies and future projections. The study found that male improvements in the CPM data pool were higher than those in the general population, while female improvements were nearly identical (results that are similar to Club Vita’s recent Canadian mortality improvements research). In this article we will discuss the considerations for using the CPM2024 base tables with different approaches to constructing improvement scales.

We look forward to helping you navigate the implications of the new tables.  If you aren’t already signed up to our mailing list, please do so using this link to ensure you are notified when future instalments of this series are released. 

Lastly, we’d love to hear from you.  Please contact us with any questions or topics you want us to cover.