Recent developments
in longevity
The S4 tables and the CMI's 2023 insurer/reinsurer survey
In the lead up to the Christmas period, the Continuous Mortality Investigation (CMI) gave us two early presents - new pensioner mortality tables and the results of its most recent survey of insurers/reinsurers. This briefing provides a summary of these publications, including key takeaways for pension scheme trustees and stakeholders.
There are a few key takeaway messages for pension scheme trustees and stakeholders that are highlighted in blue throughout this briefing summarised below:
- The draft versions of the latest S4 tables were published in November 2023.
- Although the tables are in draft, consider your plan now for adopting the S4 tables. As the current loading to S3 you are using won't map directly across to the S4 tables, this is likely to include agreeing to undertake a mortality analysis to set the base table loading.
- Pension schemes should be considering how and to what extent to make an allowance for the consequences of the coronavirus pandemic on life expectancy and liabilities. This may include referring to the impact allowed for by insurance and reinsurance companies.
Draft S4 tables
Released November 2023
According to analysis by The Pensions Regulator, around 85% of UK defined benefit pension schemes use the Self-Administered Pension Scheme (SAPS) mortality base tables (published by the CMI) to calculate the life expectancy of members in their scheme.
It is therefore a big deal when the CMI announces the release of new tables, which it did in November 2023, announcing the release of draft versions of the fourth edition of the SAPS tables - appropriately known as the 'S4' tables. Final versions of the tables are expected to be published in February 2024.
The S4 tables include the latest available data on the lives and deaths of UK defined benefit (DB) pension scheme members for data up to the end of 2019. Data for 2020 and 2021 is excluded from S4 due to the high number of deaths in these years resulting from the coronavirus pandemic (discussed later in this briefing).
A mortality base table is a way to calculate how long an individual is expected to live. Mortality base tables include mortality rates, which are probabilities of surviving over a one-year period at a specific age. Mortality base tables are typically constructed using information on the lives and deaths of a large group of relatively similar individuals, such as males/females and UK DB pension scheme members.
Socio-economic contribution to tables
The new S4 tables bring the socio-economic status of members (using a combination of pension in payment amount and postcode) into the spotlight. Previous research by the CMI and other organisations has linked where members live and their income in retirement to how long members are expected to live in retirement – higher income and living in a less deprived area typically result in living longer in retirement.
While pension scheme trustees and consultants have been using socio-economic information to set mortality assumptions for several years, this new development emphasises how important socio-economic analysis has become for calculating pension scheme liabilities. This new development is likely to extend the use of socio-economic analysis to smaller pension consultants and pension schemes and will also assist the pension scheme industry to verify the socio-economic models currently in use.
If your scheme has undertaken a socio-economic analysis in the past when setting mortality assumptions, we’re not expecting that undertaking an updated analysis using the S4 tables will result in a material change in liabilities. However, if you undertake a first analysis when adopting the S4 tables, you may see more material changes in liabilities.
Adopting the S4 tables
Most pension schemes which are using SAPS tables will currently be using the third edition of the tables, S3. The S3 tables were released in December 2018, and are therefore becoming a bit outdated (particularly in the context of developments since the coronavirus pandemic, which I discuss in more detail later). Most pension schemes will therefore be looking to adopt the S4 tables, but what should you be thinking about when deciding which of the 42 S4 tables to adopt for your scheme?
The most logical starting point is the S4 table that's equivalent to the current S3 table you're using - for most pension schemes this will probably be the S4PXA tables.
Let's say that at the last actuarial valuation, your pension scheme funded assuming 105% of S3PXA (with 105% representing a 5% increase in the probability of death at all ages compared to the unadjusted S3PXA tables). In a straightforward world, you would then move to funding assuming 105% of S4PXA. Unfortunately, adopting the S4 tables isn't as asimple as retaining the same loading as last valuation due to changes in the dataset underlying the S4 tables (and possibly also due to changes in the underlying membership of your pension scheme).
When adopting a new base table, such as the S4 tables, we advise our clients to undertake some type of mortality analysis to set the right loading, which our longevity experts at Barnett Waddingham can help you do.
Further information on the services we provide can be found on our Longevity analyses for DB pensions web page. In summary, the type and frequency of analysis that we typically recommend are:
We only typically recommend mortality experience analyses are undertaken for medium and large schemes, as the relatively low numbers of deaths for smaller schemes makes the mortality experience analysis unreliable. As socio-economic status also doesn’t tend to change significantly from valuation to valuation, there’s limited gain to be had from undertaking a socio-economic analysis for a smaller scheme using the same mortality base table.
The final S4 tables are expected in February 2024. Given there may be changes between the draft and final S4 tables, I would typically wait until the final S4 tables have been published before doing any scheme-specific analysis. In the meantime, pension scheme trustees can get ready for the release of these tables by agreeing with their advisers/administrators and sponsoring employers the scope of any analysis that will be undertaken to move to the S4 tables.
Why is having accurate mortality assumptions important?
Setting an accurate base table loading is more important than ever in the current environment:
Many pension schemes have been gradually de-risking their investment strategies over time as memberships mature and pension scheme funding improves. This de-risking trend has been accelerated by the September 2022 ’mini-budget‘, which led to a sharp rise in gilt yields and a general improvement in pension scheme funding. As investment risk has decreased, longevity risk (the risk associated with members living longer than expected) has increased as a proportion of pension schemes’ total risk. This is increasingly placing longevity risk in the spotlight as a key risk to manage – the first step of which is to accurately calculate member life expectancy based on the information we collect on members (their socio-economic profile and numbers of actual deaths in the pension scheme).
Another result of the improvement in pension scheme funding positions due to the mini-budget is that many pension schemes are either undergoing or considering a risk transaction such as a buy-in/out. A key assumption that determines the buy-in/out price is the mortality assumption adopted by the insurance company for the pension scheme, and so having an accurate idea of the likely member life expectancy the insurance company might calculate will help to decrease the chances of any unexpected surprises when insurer quotes are received. Mortality analyses can also be a helpful early indicator of wider data issues that can be addressed prior to a risk transaction, for example if a significant proportion of postcodes are missing.
Allowance for the pandemic in the S4 tables
No piece on longevity would be complete without a mention of the coronavirus pandemic – and rightly so as it has been a hugely disruptive event for calculating life expectancy.
This briefing note from Will Rice (published in March 2023, but still relevant today) highlights what’s been happening to deaths in the general population since the rollout of the Covid-19 vaccination programme in 2021. In summary, deaths have remained elevated primarily as a result of pressures on the NHS and due to Covid-19 still circulating in the general population. As pressures on the NHS seem unlikely to be solved anytime soon (even if additional funding was made available to the NHS, it would take several years for the funding to materialise into improved NHS services) and Covid-19 is likely to remain circulating in the general population, pension scheme trustees will therefore need to consider how to allow for these factors when setting life expectancy.
The S4 tables only include data up to the end of 2019, and so include no data since the start of 2020. The CMI has therefore firmly placed the decision for how to reflect the pandemic on to pension scheme trustees – should this be done as an increase in the probability of death today, or allowed to come through more slowly in future years (the approach currently adopted by the latest version of the CMI Mortality Projections Model, CMI_2022).
My view is that allowing for the coronavirus pandemic in either the base table loading, or the future mortality projection assumption is reasonable for a pension scheme, which is typically most concerned with a total liability figure rather than an exact cashflow profile. The most important thing to watch out for is not to under or over-count the allowance made, and to be clear where any allowances for the pandemic are being made in the mortality assumptions.
One consideration for setting the impact of the pandemic on life expectancy, looking at the approach taken by insurance and reinsurance companies, is discussed further in the next section.
CMI insurer/reinsurer survey
Most recently published in June 2022 and September 2023
Given the uncertainty involved with deciding on the impact of the pandemic, it is informative to consider the approach adopted by other organisations. Considering the approach adopted by insurance and reinsurance companies is helpful as buy-in/out pricing becomes an increasingly important part of pension scheme funding.
For the last few years, the CMI has undertaken a survey of the mortality assumptions used by insurance and reinsurance companies. These surveys can provide clues on how buy-in/out pricing is changing over time.
The two most recent surveys asked respondents how much they think the coronavirus pandemic has impacted on life expectancy:
-
2022 survey (published in June 2022): average life expectancy reduction of 0.3%
-
2023 survey (published in September 2023): average life expectancy reduction of 1.7%
The survey in 2022 was issued at a time when the industry was still being relatively cautious, likely leading to the relatively small life expectancy fall of 0.3%. Since the 2022 survey, a clearer negative trend for life expectancy has emerged, likely leading to the larger life expectancy fall of 1.7%.
In general, I would expect this fall in life expectancy would materialise in an improvement in buy-in/out pricing for pension schemes. However, this improvement in pricing (of approximately 1.7% minus 0.3% = 1.4%) is a relatively small factor in the overall setting of pricing (such as investment returns and supply/demand).
It will be interesting to see the responses to this question in the next survey in 2024 – will the reduction in life expectancy be more than 1.7% if insurers/reinsurers are more certain of the trend in mortality emerging after the end of the pandemic?