What’s COVID-19 done to term assurance experience?
The CMI Assurances Committee reports regularly on the claims experience of policyholders with term assurances issued by UK life insurers. Our most recent analysis, released in Working Paper 162 considers the experience in 2020 and the initial impact of COVID-19.
What has the Committee considered in Working Paper 162?
We consider experience for three benefit types separately: mortality, accelerated critical illness and stand-alone critical illness. Our focus in the paper (and in this blog) is on the larger two benefit types – mortality and accelerated critical illness.
We compare experience against the “16” Series term assurance tables. More information on the derivation of these tables can be found in our previous blog.
In the main we compare 2020 experience against that seen over the combined period 2016-2019. We also consider 2020 experience by month, vs 2019, in order to assess the impact of COVID-19 on experience.
What did you see in the 2020 experience?
For mortality we see that overall lives-weighted experience is around 4% higher overall than is expected by the “16” Series term mortality (T16) tables. Compared to 2016-2019, the experience in 2020 is heavier for males and female non-smokers, but slightly lighter for female smokers.
Since we're only considering a single year, it’s unsurprising that there’s some variation in the results, but the age shape of the experience looks broadly similar to the T16 tables. The exception is at older ages where experience is relatively heavier, suggesting that older ages in the term mortality dataset were impacted more by COVID-19.
For accelerated critical illness overall the experience in 2020 is 14% lower than expected by the “16” Series (AC16) tables. As with mortality, the age shape is broadly similar to that of the underlying table, albeit at a lower overall level. This suggests the impact of the COVID-19 pandemic was fairly consistent across all ages. An exception is male smokers where experience appears too erratic to make any meaningful comment.
What did the results by amount show?
Unlike lives – which was heavier - the amounts-weighted experience in 2020 is similar to the earlier period. Consequently the difference between amounts-weighted and lives-weighted results is larger in 2020 compared to 2016-2019.
The Committee investigated this further by considering experience by sum assured band. As with previous analyses, mortality experience reduces with increasing sums assured for both periods, but a noticeably steeper gradient is seen in 2020 compared to the earlier period. The experience in the lowest band, £0-£25,000, is particularly heavy in 2020, whereas the experience in the highest band, £250,001+ is substantially lighter than in 2016-2019.
A different pattern is seen for accelerated critical illness. Here the lives-weighted experience in 2020 is lighter than 2016-2019 for all subsets, and the amounts experience is broadly similar to the lives experience across both periods.
What other analyses did you perform?
We considered several one-way analyses of the experience in 2020, compared with 2016-2019, including experience by distribution channel, product type and commencement year.
Interestingly, for accelerated critical illness, we see that the latest commencement year group, 2016 – 2020, has the heaviest experience in the 2020 dataset. It’s not clear what the underlying driver of this is, nor whether this increase will persist to later durations, but it could be indicative of a switch to less stringent levels of underwriting or increased anti-selection in these years.
What analysis did you do on the impact of the COVID-19 pandemic?
We were keen to build on indicative work in Working Paper 152 and consider 2020 experience by month, compared to 2019, in order to assess the impact of COVID-19 on experience.
For mortality we see a large spike in April 2020 coinciding with the first wave of deaths in the general population. We also see evidence of increased mortality towards the end of the year, coinciding with the second wave.
The spike in experience in April 2020 appears to be primarily driven by a large increase in male non-smoker experience, and female non-smokers to a lesser extent. The spike is less evident for smokers. We don’t think there is a clear reason for this but it could suggest mortality from COVID-19 is similar for non-smokers and smokers. Since we compare experience with smoker-specific tables, non-smoker experience is compared with a lower baseline than smokers so that could be why we observe heavier relative experience for non-smokers compared to smokers.
Overall if we estimate excess claims in 2020 (vs 2019) we see relative excess mortality of +9% across ages 30-75. This compares to a corresponding figure of +12% for that age group in the England and Wales population. It suggests that COVID-19 had less impact on policyholders of term mortality assurances than it did on the general population, but (as noted in the paper) there are some caveats to this comparison and the impact varies by gender.
For accelerated critical illness we see that experience in 2020 is lighter than expected in all months. It is lowest in May 2020 and then increases, monotonically, back to 2019 levels. The impacts are similar for both genders and smoker statuses.
It’s of course highly unlikely that this drop represents a genuine fall in experience. More likely is that it reflects a displacement of claims caused, for example, by disruption to routine medical and screening services. If this proves correct, then we would expect more claims to emerge over time, but there is so far no evidence of these displaced claims arising in late 2020.
What is the Committee working on next?
In addition to our regular updates on experience the Committee is looking to undertake an additional analysis in 2022. The Committee is discussing four areas of analysis:
- Rated lives;
- Cause of claim for accelerated critical illness,
- IMD analysis, and
- Persistency analysis
The Committee is assessing the feasibility of each of these, based on data already held and whether our data contributors support such additional analysis. We currently believe a rated lives or cause of claim analysis would be a feasible option this year, but we’d very much welcome views on the relative benefits of these and any other initiatives that users would find valuable.