Steve Bale, Senior Actuary at Munich Re and a member of the CMI’s Mortality Projection Committee, blogs on the questions facing Life insurers as they consider the impact of COVID-19 on future claims on their insured lives. The views expressed in this blog are his own.
The COVID-19 pandemic has led to daily reports of numbers of infected cases and resulting deaths, and an attention to identifying trends amongst the noise; are deaths related to COVID-19 stabilising, falling or are we heading for a second wave of infections?
Life insurers are considering the impact on future claims on their insured lives. The key questions will include:
- What allowance should be made for extra claims cost in pricing and reporting?
- How might current base mortality assumptions be rolled forward or adjusted given mortality effects in 2020 (and potentially future years)?
- How might future trends in mortality and morbidity change in the future for the residual lives insured?
We would expect to see higher claims on mortality and morbidity-related insurance policies, although the impact of COVID-19 on an underwritten and generally more affluent sub-population of the general population is not immediately clear at this early stage. There is also uncertainty about the impact of indirect claims later down the line resulting from medical diagnoses and treatment being postponed. For example, University College London Cancer Institute has estimated that there has been a 76% decrease in urgent cancer diagnosis referrals from GPs and a 60% reduction in chemotherapy appointments held in the early stages of the pandemic lockdown and this might be expected to result in additional deaths in 2020 and 2021.
In absence of credible data on COVID-19 related deaths, insurers could consider population mortality data as an alternative to insured lives data and how it might help inform expected views for insured lives.
Population data provides early insights into mortality trends
Unlike insured lives experience which typically has a reporting lag after the claim event, death data from the Office for National Statistics (ONS) can help inform what is happening at a population level. In many ways we have benefited for a number of years from the ONS publishing weekly death registrations, with splits by age and gender, with typically a two week delay. Usually deaths have to be registered in England & Wales within 5 days of occurrence, meaning that we can track death trends across a calendar year and consider seasonal trends as well as early indications as to how mortality experience might develop over a calendar year. As well as measuring COVID-19 identified deaths, the ONS also report on total population deaths, providing an insight on deaths which may be COVID-19 related but not identified as such, or indirect deaths. Looking at total excess deaths can provide a more robust view on potential impacts.
It could be said that we have been spoilt by having free access to this up-to-date dataset, and in the midst of an extreme international pandemic the ONS’s information has suddenly become a focus of attention for early indicators.
Increased analysis in uncertain times
In addition, since the pandemic started the ONS has amended the weekly reporting so that deaths where COVID-19 was identified as having been associated either through a positive test or doctor diagnosis has been recorded. The CMI’s Mortality Projection Committee has identified that as at mid-May, most of all of the excess deaths for March through to May can be attributed to COVID-19. It will be interesting to consider how this progresses over the remainder of the year and in 2021 as the pandemic progresses and other indirect deaths occur.
In addition, the ONS have been expanding their cause of death analysis so that we can track trends in death cause and other contributing risk information. Typically this data is published with up to a year delay, but since the pandemic began the ONS has reprioritised and we now have access to the previous month’s death information within two weeks of the month end.
This cause of death analysis is particularly useful as it contains information about pre-existing conditions held by the deceased at time of death where COVID-19 is associated, meaning that we can assess the key risk factors by age and gender that might be influencing mortality. For example, at time of writing just under 10% of COVID-19 associated deaths aged over 65 had no pre-existing condition.
There does remain the challenge however for (re)insurers when considering what an insured population might be affected by COVID-19 look like in terms of pre-existing conditions assessed and managed by the underwriting process. In addition, many of the ONS analyses are 1- or 2-way, whereas actuaries are more familiar with assessing multi-factorial datasets.
A collaborative opportunity?
The CMI have already increased their monitoring of population mortality to provide weekly summaries of the weekly deaths produced by the ONS, as well as commentary on the excess deaths over those expected for 2020. Attention will in turn to what it might mean for CMI and other model and analysis outputs. There may also be the opportunity for the IFoA/CMI to work with the ONS and combine the strength of both organisations data and strength of analytical capability to maximise learnings from this pandemic.