27 April 2020
In this blog, Matthew Edwards, Editor of the IFoA Longevity Bulletin, talks about life expectancy and the contribution actuaries can make during the COVID-19 crisis.
Are people dying ‘of’ COVID-19 or ‘with’ COVID-19? What sounds like a question of linguistic pedantry has become a vital question, the answer to which has great implications for many countries’ approaches to the pandemic. This is one of the vital COVID-19 questions that actuaries are uniquely well-placed to advise on.
Comment and speculation in the media about ‘of’ or ‘with’ has been growing, fuelled in part by a comment by Professor Neil Ferguson that ‘as many as half to two-thirds of deaths from coronavirus might have happened this year anyway, because most fatalities were among people at the end of their lives or with other health conditions’. In other words, they died coincidentally ‘with’ the coronavirus in their bodies.
This seems not only wrong to me – as we’ll discuss – but could also lead to a callous attitude regarding sufferers of COVID-19. ‘Why should we care, they were about to die anyway?’ And this attitude can then become a rationale to ignore social distancing. The stance of ‘they’re dying ‘with’ not ‘of’’ COVID-19 is the central argument used by some commentators for an easing of the lockdown as soon as possible.
This is a debate that should be had properly, with full consideration of the arguments in all directions, but the idea that a large proportion – let alone the majority – of COVID-19 victims were going to die soon anyway is completely wrong.
How does actuarial expertise help here? There are two important things we can contribute (quite apart from our contribution to pandemic modelling in general):
- We are very well placed to see through data confounding issues;
- We are uniquely placed to advise on life expectancy.
The first point relates to why the attitude of ‘they were going to die soon anyway’ has prospered. Many people have seen statistics noting that large proportions of the deaths were people who had an existing ‘condition’, and added two and two to make an imaginary number. But we can appreciate that the deaths are weighted towards the old, and most of the elderly have one or more ‘conditions’ – which may be serious (chronic obstructive pulmonary disorder (COPD), for instance), or not (hypertension, a past heart attack, obesity – not so serious, as we see below). Saying ‘they had an existing condition’ as if that ‘exculpates’ the coronavirus is like saying ‘they are old, so they died’. Our actuarial ability to disaggregate data and not be misled by confounding factors is a valuable skill.
The second point, regarding life expectancy, is fundamental. Few people outside the actuarial and medical professions appreciate how high your life expectancy can be even if you’re old and not in good health. A typical lay misunderstanding runs along the lines of ‘Life expectancy at birth is 80 years or so, so if you’re in your 80s you’ll die in the next couple of years’.
But actuaries have been reserving for and pricing normal annuities for many years, and many of us have also been doing so for impaired lives, helping them get better annuity rates. We can take the sort of ratebook or underwriting engine underlying much of the UK’s annuity business and consider what numbers we see for apparently unhealthy people.
Let’s consider diabetic obese male smokers as a reference group here. Not great role models, and if you read about an old diabetic obese male smoker dying from (with?) COVID-19 it feels almost intuitive to think ‘the carbs and fags got him’, not ‘the coronavirus got him’.
But the life expectancy of these non-athletes is more ‘athletic’ than you’d think. Even assuming no mortality improvements, to stack the game against them more, a 70-year old of diabetic obese male smoker has a life expectancy in the region of 8-9 years; an 80-year old has a life expectancy of 4-5 years. If we take away the diabetes and instead assume COPD, or a heart attack a few years ago, we get roughly the same results.
Yes, you say, but life expectancy is an average and there’s a lot of variation around that average. That’s true, of course, but the odds of our reference group surviving the year are good. Only around 1-in-20 of such 70-year olds, or 1-in-7 of such 80-year olds, would be expected to die within twelve months.
These are examples of people who have a respectable number of years to look forward to; they are not ‘the walking dead’, and if they die with COVID-19 raging in their bodies, we can accurately say that they died of COVID-19, and were probably not going to die this year in any case.
Using our skills in this way, to inform the debate where we are uniquely placed to do so, is just what we should be doing.
The main points of this blog are expanded on in an article by Matthew Edwards and Stuart McDonald in the May 2020 edition of The Actuary.