Mortality Monitor Q2 2019

What does the latest population data tell us about emerging trends?

The last few years have seen relatively high death rates in the UK. However 2019 has, to date, seen relatively few deaths. We explore below the latest data, and consider the implications for pension schemes.


Statistics on the numbers of deaths at a population level can provide a useful ‘temperature check’ on the emerging trends in longevity. Given that the CMI’s mortality improvements model is calibrated to England & Wales population data, we can also use published information to assess the potential impact on next year’s model in advance of publication.

Recent years have seen a marked increase in deaths at a population level. There has been some debate around what has been driving this increase, with various commentators suggesting factors such as the cumulative effects of heavy winters and virulent flu, the rise of Alzheimer’s disease, improvements in circulatory disease having run their course, or the impact of austerity based squeezes on health and social care budgets.

This recent heavy experience has also had an impact on projections for future improvements in mortality rates, with successive updates typically resulting in reductions in the assumed rates of improvement, and so liabilities. Indeed the most recent version of the CMI’s mortality improvements model, CMI_2018, could lead to liabilities that were around 2.5% lower than under the previous version (although around half of this was due to a technical change in the model’s ‘core’ calibration – see my previous blog for details).

There are signs that 2019 is (so far) turning out somewhat different to recent years though. It began with a very mild winter, and the impact of flu on mortality was limited. There was also a noticeable absence of “NHS in crisis” media headlines. We can now start to look ahead to consider what the rest of 2019 may hold, and the implications for pension schemes.

The latest data on population experience

We keep a close eye on the weekly deaths statistics for England & Wales, as published by the ONS. We can see how the weekly deaths over the first half of 2019 are lower than any of the preceding 5 years.

In considering weekly death numbers, it is important to also allow for the expected impact of the aging population.  If mortality rates were to remain unchanged, we would still expect the number of deaths to increase by 2-3% from one year to the next. For deaths to fall we would therefore have had to experience relatively strong improvements (reductions) in mortality rates to counteract the impact of ageing.

The CMI have now published the latest version of their quarterly Mortality Monitor, covering their analysis of the ONS data for the first half of 2019 (to the end of week 26, 28 June 2019). This analysis focuses on standardised mortality rates (SMR), which control for changes in age profile of the population over time.  These rates are a useful way to quickly compare changes in mortality rates over time, although clearly some of the finer detail of age based trends can be lost as a result of standardisation.

The main headlines from this edition are as follows:

  • The annual average SMR (the rolling average of weekly SMRs over successive 53 week periods) has fallen steadily from mid 2018 to mid 2019;
  • The quarterly average SMR (the rolling average of weekly SMRs over successive 13 week periods) has continued to be relatively light; and
  • Cumulative SMR in 2019 (the SMR over the period to the end of week 26) has remained well below the average of the previous 10 years, and is similar to that seen in the corresponding period of 2014 (which saw the lowest mortality rates over the last 10 years).

What this means for pension schemes

The CMI’s mortality improvement model is the standard way of setting mortality improvements. The model is typically calibrated to England & Wales population data. The model includes the ability to adjust many of the parameters to control both how the model fits to historical data and the way it projects future improvements to a (user defined) long term rate. It also includes so called ‘core’ settings, which, while not intended to necessarily represent appropriate values for pension schemes, nonetheless are a useful way of comparing different variants. 

The most recent version of the model, CMI_2018, introduced a change to the core value of the Sκ smoothing parameter which controls the sensitivity of the model to more recent data, reducing it from 7.5 to 7. This had the effect of increasing the sensitivity of the core parameterisation of the model to recent data.

The chart below looks at the range of potential values for annuity values that we may see in CMI_2019 depending on how the rest of the year turns out.

We also show corresponding values for the periods underpinning the previous variants of the CMI model, which show the downwards trend over the last few years. In doing so we show the published parameterisation.   

Note: The annuity values shown are for men at age 65, based on the S3PMA base table, with a net interest rate of 0% p.a. and assuming a long term rate of 1.5% p.a. in each case. Calculations were carried out using the CMI_2018 model, and the E&W population data included with that model. CMI_2019 values are based on fitting the model to 1979 to 2019, where the level of deaths in 2019 have been estimated using improvements of 3.6%, 4.8% and 6.0%, and assuming exposures in 2019 are equal to those in 2018.  The calculation date reflects the data used, so the CMI_2019 estimates use a calculation date of 1 January 2020, the CMI_2018 value uses a calculation date of 1 January 2019 etc.

We can see how it is likely that there will be at least some ‘bounce back’ in annuity values (and so liabilities) in the published CMI_2019 model (assuming no further changes to core settings).

Schemes which have been adopting core values without adjustment will therefore see an increase in liabilities, potentially to values close to those seen under CMI_2017.

Looking forward

As you’re no doubt aware, July has seen record breaking heat. And of course time will tell what winter brings (given the trend towards more extreme weather events). In addition, Australia saw an early start to their flu season – although there are signs that it may finish earlier too (and what happens in Australia is typically a good indicator of what to expect in our next flu season in the UK). Therefore there remains plenty of scope for things to change in the remaining half of the year.

And looking further ahead, we need to consider whether any ‘bounce back’ in 2019 is expected to be a blip or a return towards the high levels of mortality improvements seen through much of the 2000s.

We will of course be continuing to monitor emerging experience over the weeks and months ahead.

This report complies with the requirements of Technical Actuarial Standard 100 as effective from 1 July 2017.

Share this article:

Get in touch

If you wish to find out more please don't hesitate to get in touch

Icon/Arrow/UpIcon/Pin/Calander12Icon/Close/blackIcon/Social/FacebookIcon/Social/LinkedinIcon/MinusIcon/PinIcon/ExpandIcon/QuoteIcon/Website-greenIcon/Website/grey