How long will people born in 2015 actually live?

The answer to this question depends on being able to determine the probable life expectancy of a birth cohort, rather than the cohort’s life expectancy at birth. The difference between these two numbers is that life expectancy at birth (LEB) isn’t a prediction: it’s a statistical fact, that is, it’s a statement of the mean number of years that will be lived by members of the cohort if the current age-specific mortality rates in the population as a whole remain steady over the cohort’s entire lifespan. Of course to the extent that age-specific mortality rates change over that time, life expectancy at birth won’t reflect the actual life expectancy of the cohort.

To give a concrete example, LEB in the US was 47 in 1900, but it’s certain that the actual average life span of people born in the US in 1900 ended up being quite a bit higher, because age specific mortality rates have dropped pretty much continually since then (they are currently dropping most sharply among the oldest members of the population). But how much higher?

If one is trying to predict how long the average American born today will live — which, for practical purposes, is a much more important number than LEB — how would one do it? Did people in 1900 end up living 10% longer than their LEB? 15%? More? And whatever the spread between LEB and actual life expectancy was, how likely is it to be replicated for people born in 2015? (In the developed world, the increase in LEB has been remarkably steady, with exceptions for a world war or two, for nearly two centuries now).

Anyway, a 15% increase between LEB and actual life expectancy would mean the average American born today will live to be 91, which probably means that in a few decades Zombie Robert Samuelson will be arguing that the social security retirement age should be raised to 83.