The CMI Annuities Committee has released two working papers based on data for pension annuities in payment in 2011-2014:
- Working Paper 101 (July 2017) contained initial analysis of the data, to accompany the “all offices” results; and
- Working Paper 112 (November 2018) contained additional analysis of the data.
We have also been collecting and processing data for more recent years and are aiming to release “all offices” results for the years 2011 to 2016 in January 2019.
We wanted to highlight to users that the more recent data has reduced our confidence in the 2011-2014 data underlying the previous working papers:
- Several insurance companies have resubmitted data for the earlier years in addition to supplying data for more recent years. This is not unusual; in particular, resubmissions may include a small volume of additional deaths that had not been reported and processed for the first submission.
- However, several of the insurers that resubmitted 2011-2014 data had undertaken data cleansing exercises and/or systems upgrades between their submissions. This included some of the largest data contributors to the Annuities investigation and, in combination, these appear to have a more significant impact on the results.
The provisional impact on the all-ages, all-durations results for 2011-2014 for all products combined are:
- Males: an increase from 93.9% of PML08 to 94.1% (lives-weighted) but a decrease from 93.5% of PMA08 to 92.3% (amounts-weighted).
- Females: an increase from 93.5% of PFL08 to 94.9% (lives-weighted) and an increase from 95.0% of PFA08 to 95.8% (amounts-weighted).
Larger differences emerge at a more granular level. These arise from a combination of factors:
- The previous dataset was “incomplete”; in particular, 2014 results were based on data from fewer offices (see Chart 4D of Working Paper 101) – we now have much of that data.
- Two insurers were able to submit data divided by product type that had previously been categorised as “unknown”.
- Two insurers are now uncertain of the reliability of the segregation of individual annuities between “internal vesting” and “external” in their previous submissions. Both have confirmed that they regard the newer data as more reliable than their previous submissions.
- One insurer is now uncertain of the reliability of the data they previously submitted for second lives (i.e. widows and widowers).
Consequently, we now have less confidence in the data underlying the results and analysis in Working Papers 101 and 112, so those using these papers should consider how much reliance to place on them.
As noted above, the Committee was aiming to release “all offices” results for the years 2011 to 2016 in January. We will continue to aim for this and, consequently, we do not propose reissuing either of the existing working papers. However, as a matter of priority, the Secretariat will seek to confirm their understanding of aspects of individual data contributor’s data and results – for example, the reliability of the separation of their data between internal and external – both where the firm has resubmitted data, and where they have not.
If you have any questions regarding this update, please email firstname.lastname@example.org; however please recognise that we may not be able to provide any additional detail until we have completed the analysis necessary for the next paper.