The new version of the CMI model (CMI_2020) was published in March 2021. In this article, we take a look at the headline results of the model, and consider how the new functionality that has been added may be used to update longevity projections in light of the Covid-19 pandemic.

The CMI Mortality Projections Model is widely used across the insurance and reinsurance industry to project longevity improvements. The new version introduces a change to the model to address the issue of the exceptional 2020 experience due to the Covid-19 pandemic being unrepresentative of underlying trends. Users can now choose the amount of weight put on data for individual years. The core model places no weight on 2020 data but full weight on data for all other years.

Impact on life expectancy

Alongside the latest version of the model, the CMI has released a working paper covering the impact of moving from the previous version of the model (CMI_2019) to CMI_2020. The working paper illustrates that adopting CMI_2020 in place of CMI_2019 (and assuming default parameters are used in each case) could result in a fall in projected life expectancies. This projected fall is around 4 weeks for males age 65 and 1 week for females age 65. It is important that we consider what these figures really mean.

What does this impact represent?

The most important point to note here is that this decrease in illustrated life expectancy is not an impact of Covid-19. Reflecting 2020 data in CMI_2020 in the usual way would significantly distort the projections, hence why CMI_2020 (using default parameters) places no weight on 2020 data.

So, why is there a change in projected life expectancy in the new model if we are not taking account of 2020 data?

A key driver of this change is the concept of the transition from short term improvements to long term improvements. Short term improvements are typically low, reflecting the low improvements observed in the latter half of the 2010s whereas long term improvements are typically higher. Each new version of the CMI mortality projections model effectively moves the long-term further out by a year. Therefore, this delays the transition to higher rates of improvement by a year. This shift is then a key driver of the decrease in life expectancy projected by CMI_2020.

Using the flexibility of the CMI_2020 model

As mentioned above, CMI_2020 provides a mechanism to specify the weighting given to data in each year.

There are several ways in which users could evolve their trend assumptions. These include:

• Continue to use an earlier version of the model (and existing parameterisation). This approach reflects a view that 2020 data tells us nothing new about how mortality will evolve. This may be used as a holding assumption which is adopted while the longer-term impacts of Covid-19 on mortality trends are being assessed.
• Adopt the core version of CMI_2020 model. This approach is consistent with the above in the view that 2020 data is not helpful in updating views on mortality trends. However, this does take the view that 2020 experience has pushed out the transition to higher long-term improvements by a year.
• Adopt:
  • CMI_2020 with some weight to 2020 data;
  • CMI_2020 with some other amendment (e.g. use of the A parameter); or
  • Retain the existing CMI model but amend it in some way (e.g. use of the A parameter).

This is a non-exhaustive list of possible approaches which may reflect the user’s view of the short-medium term future impact of Covid-19. The key challenge of implementing approaches of this kind is that the user will have to form a view on the impact of Covid-19 on future mortality trends outside the CMI mortality projections model. The parameters in the model will then have to be set to reflect those views.

It is important to note that although a default parametrisation of CMI_2020 is given, it is not intended to be in any way “recommended” by the CMI. The value of the CMI mortality projections model is that it provides a common framework within which users can express their views (by using various parameters within the model).

Summary

As noted above there are a wide range of possible approaches to updating mortality projection assumptions. These approaches range from making no amendment (which many will adopt while forming a robust view) to much more sophisticated updates. In practice we may expect many users will have adopted approaches such as the former over the course of 2020 but will move to more sophisticated assessments of the longer-term impacts during 2021.

The Hymans Robertson team has been researching and quantifying the longer-term mortality and morbidity impacts of the pandemic and so is well placed to help in this area. Please get in touch if you would like to find out more.