Why do we underestimate how long we might live?

People tend to underestimate how long they are going to live. Why is this? One contributing factor is that the life expectancy “facts” we usually see or hear are themselves underestimates.

People tend to underestimate how long they will live. When asked What age do you think you will live to? New Zealand men underestimated on average by over five years and women by over seven years, compared to realistic expectations of life from tables which allow for current age, gender and future longevity improvements*. This is a similar pattern to that found in American and British surveys**.

Why is this? The reasons boil down to not understanding longevity trends. The New Zealand study suggested that:

  • People take cues from the contemporary pattern of deaths in the population without taking likely future longevity improvement into account.
  • When asked What age do you think you will live to? a person’s first thoughts are often their parents’ or grandparents’ experience, ignoring that each generation is on average living longer.
  • The average gap between declared and realistic expectation conceals substantial variation. The odds of underestimating are twice as high for women compared to men, but men are more likely to be extremely over-pessimistic and over-optimistic; both extremes being associated with low financial knowledge.
  • Younger people are more likely to underestimate than older people, because the effect of ignoring future longevity improvement is greater.

Does this matter? Yes. If people underestimate how long they will live they suffer longevity risk in their retirement planning – the risk that their money runs out sooner than expected. Longevity risk also exists in policy making. If there is persistent underestimation of population average lifespans, then public pensions cost more than expected and the provision of services such as long-term care may not match demand.

So what can we do? Clearly, realistic indicators for likely lifespans are needed, and they need to be referenced consistently to avoid confusion. Which is why it’s disappointing to read this in a Press Release from Public Health England (PHE):

“According to PHE’s report on recent trends in life expectancy at older ages:

  • men can now expect to live for a further 19 years at age 65, 12 years at 75, 6 years at 85 and 3 years at 95
  • women can expect to live for a further 21 years at age 65, 13 years at 75, 7 years at 85, and 3 years at 95″

This uses a period life expectancy indicator as if it gives the lifespan we should expect. It doesn’t – as this blog post explains.

PHE correctly point out towards the end of their report that the period life expectancy indicator they use

“… reflects mortality among those living in the area in each time period…. It is not therefore the number of years a person in the area in each time period could actually expect to live…” (page 21, emphasis added)

The period indicators used by PHE are appropriate to compare summary mortality measures at one period in time across different regions of England, and to compare across time periods. So they are used correctly for the purposes of PHE’s report.  However, they are not realistic indicators for likely lifespans so shouldn’t be used for the “can expect to live” message in the headline without further explanation.

Cohort life expectancy indicators are a more realistic measure of how long someone is actually expected to live, because cohort indicators take into account likely future mortality trends. Cohort indicators are available from the same source used in the PHE report (ONS Principal Projection for England, 2014). Cohort life expectancy for a man aged 65 in 2014 is 21.4 years, compared to the period headline figure of 19 years. A 2.4 year underestimation for today’s 65-year-olds  may not sound much, but the younger you are the more longevity improvements will kick in and widen that gap – see this chart.

So to avoid longevity risk men shouldn’t be reading “19 years” as the headline for how long they might live after age 65. They should understand “at least 21 years – and more the younger you are”. Most of the average 5 to 7 year gap between expectations and likely reality can be explained by expectations being influenced by headline data which has this level of underestimation built in.

For consistent messaging of how long we can expect to live, it would help if every institution which used the phrase “can expect to live to” had an alarm bell which prompted a check on the real meaning of the data and if necessary the addition of a rider:

“assuming there are no further gains in longevity”.

Or even a full disclaimer:

“For a more realistic estimate of how long you might live allowing for anticipated gains in longevity, add a couple of years if you are aged 65 and another year or so for each decade by which you are younger than 65“.


* A summary of these survey results is available here.
** International comparisons are covered in the IPS chapter above, and also this article in the Journal of Population Ageing.

Leave a Reply