During 2020 and 2021, the UK saw very high numbers of deaths due to the direct impact of the COVID-19 pandemic. Despite comparatively high levels of COVID-19 infection continuing to circulate, death rates in 2022 have so far stabilised to be roughly in line with what was seen in 2019.
To estimate the likely cost of providing benefits to members, pension schemes need to take a view on what might happen to mortality rates over the coming years - this view will be expressed via a mortality projections model. For a model to produce robust short and medium-term projections, it is important that it contains reasonable and up-to-date data - which is difficult given recent experience. The pandemic years are not only years that cannot be reliably used to predict future years' deaths but are also a period where we have had more than two standard years of change. Among other things, many people's work patterns are unrecognisable from before the-pandemic, and the way we access healthcare systems has changed permanently - for example, the proportion of outpatient attendances taking place virtually was 4% in February 2020 and is expected to remain at 25% for the foreseeable future.
The impact of COVID-19 has further obscured a pattern of improvements that was already uncertain - and this is even without thinking about the impacts of non-pandemic factors such as the cost of living. All of this makes it much harder than usual to come up with a suitable mortality projection to use, and making it more important than ever to take account of data other than just recent death rates when setting a mortality basis.
Additional important signals to bear in mind when considering longevity expectations are:
• Endemic COVID - as we come out of the pandemic period, we expect COVID-19 to continue to circulate and be a significant cause of death for the foreseeable future. Spring / Summer 2022 should give a reasonable estimate of the number of COVID-19 deaths that might be seen in the UK once a steady state is reached.
• Long COVID - the Office for National Statistics (ONS) publishes regular studies looking at the prevalence of symptoms that persist long after COVID infection. While initial expectations are that the mortality impact of long COVID on a population-wide basis is likely to be relatively small, there is still considerable uncertainty as to the number of people severely affected and the knock-on impacts to their health. Increases in diagnosis of related diseases (for example heart disease) could give an indication that long COVID might have a larger impact.
• Waiting time statistics - over the last two years, waiting lists for elective care have increased significantly, and at the same time as fewer elective procedures have been carried out. Although additional funding has been promised to address the backlog, waiting lists are projected to continue to rise at least in the medium term. This means that there have been, and will continue to be, delays in receiving treatment. Assuming that the NHS is able to bring waiting lists to manageable levels over the next few years, the impact on mortality may not be great. However, monitoring waiting time statistics along with trends in diagnoses should give an early indication of the speed at which normality is reached.
• Cost of living indicators - fuel poverty and food insecurity (that is, where eating patterns are disrupted and food intake reduced due to lack of money for food) both have the potential to increase mortality in the short term, although the likely number of additional deaths is very unclear. Increases in conditions related to fuel poverty (principally circulatory and respiratory diseases) may be an indication of a material impact.
In general, risks to members of defined benefit pension schemes have been relatively well mitigated (due to having an inflation-linked pension), compared to the risks to the wider population who may not have income or assets which increase in the same way. However, very high inflation could be of greater concern for many pensioners, because in most schemes the increase is capped (and some elements of their pension may not be index-linked).
It will be important for pension schemes and the insurance industry to reach a view on the likely level of improvements in the future (and in particular, whether expectations have changed from pre-pandemic levels). While the shock of 2020 and 2021 has the potential to change the pre-pandemic trend of mortality improvement, based on prior experience we expect that it could take three to five years for a consensus to be reached. The generally accepted view is that there was a change in improvement trend in 2011, but the consensus on this (led by Aon calling out a dislocation in the longevity market) did not really start to form until 2016. Reaching a post-pandemic consensus is likely to take at least as long, partly because the pandemic makes it harder to use the latest mortality data to inform future expectations, and partly because mortality trends are inherently uncertain. Given the levels of uncertainty, bulk annuity transactions and longevity swaps remain effective forms of protection against longevity risk.
This post is funded by Aon