What would you do differently if you suddenly found out that you’ll probably live about seven years longer than you currently expect? The question is not hypothetical. Congratulations! You’ll probably live longer than you expect. Most people have heard of life expectancy statistics before, and intuitively understand that such statistics are meant to capture how long people in a given population are expected to live, on average. But how is life expectancy calculated? As it turns out, there are actually several different types of life expectancy statistics, and the one you’ve probably encountered has a different meaning than you might expect.
Period Life Expectancy
The life expectancy statistic that you’ve probably seen reported on in the news is called period life expectancy, which is usually used to describe changes in population health over time. Period life expectancy statistics are calculated using the age-specific mortality rates for a given period (usually a specific year, such as 2020). Essentially, the period life expectancy for a given year describes how long a person would be expected to live if they were to experience all of the age-specific mortality rates that prevailed in that population in that specific year. In other words, the 2020 period life expectancy at birth is calculated assuming that the cohort of children born in 2020 will grow up to experience the mortality rate of under-ones in 2020, the mortality rate of one-year olds in 2020, the mortality rate of two-year olds in 2020, the mortality rate of three-year olds in 2020, and so on. This is, to put it simply, impossible for any particular birth cohort to experience in real life, since mortality rates change over time-the 2020 birth cohort cannot live their entire lives in 2020. By the time the cohort of children born in 2020 reached their first birthday, they became subject to the 2021 mortality rate for one-year olds, not the 2020 mortality rate for one-year olds. As time goes on, the gap between the mortality rates that the 2020 birth cohort actually will actually experience and the mortality rates experienced by people of all ages in 2020 will likely grow ever-wider, due to advances in public health, medicine, economic growth, and so on.
This dynamic means that period life expectancy does not correspond to the mortality rates experienced by any particular real-world birth cohort. So why do we use it anyway? First, period life expectancy is still useful for demonstrating changes in population health over time. However, instead of comparing the average length of life of particular birth cohorts over time, what we’re doing instead is comparing a simulated life expectancy based on the age-specific mortality rates that prevailed in a given year (say, 2020) with a different simulated life expectancy based on the age-specific mortality rates that prevailed in a different year (say, 2000). Second, period life expectancy can be calculated using data that already exists, so it’s straightforward to calculate, unlike the main alternatives.
Cohort life expectancy
Cohort life expectancy represents the observed average length of life experienced by a given cohort (say, people born in 1900). This is probably what most people imagine when they hear the words “life expectancy”. For cohorts from the distant past in which everyone is dead, cohort life expectancy is straightforward to calculate, since we can simply count how many years people in the cohort lived, on average. Things get more tricky for more recent cohorts in which some people are still alive. How do we know how long those cohorts will live, on average? The short answer is that we don’t. But demographers estimate it as carefully as they can, based on past changes in observed cohort life expectancy over time.
So what’s the difference between period and cohort life expectancy?
I created the graph shown above using data from the UK Office for National Statistics (Morgan, 2019). It has several clear implications. First, since 1850 or so cohort life expectancy has substantially outpaced period life expectancy in industrialized countries like England. Since 1950, the gap in England and Wales has averaged about 8 years and stood at about 7 years in 2020. This gap reflects the fact that over this period there have been enormous improvements in public health, medicine, and average incomes which mean that for the most part, each successive generation lives longer than they would have if their cohort had experienced the age-specific mortality rates that prevailed in their birth year. Second, it is worth noting that period life expectancy is much more variable year to year than cohort life expectancy. This is because period life expectancy reflects changes in all age-specific mortality rates that prevailed in just one particular year in which particular events like a pandemic might be very influential, while cohort life expectancy is determined by age-specific mortality rates that play out over a whole lifetime, rendering those particular events comparatively less important. If you turn your attention to 1918, you can see that the 1918 influenza pandemic caused a sharp 7-year decline in period life expectancy, as you might expect, while cohort life expectancy actually increased slightly in 1918, which might be a surprise. Why did cohort life expectancy continue to rise in 1918 despite the pandemic? Infant and child mortality for the 1918 birth cohort was not increased enough by the pandemic to offset other long-term forces that were causing the consistent decline in infant and child mortality seen over the 20th century.
Why does this matter?
Our ideas about life expectancy influence many of our most important life plans, so it’s important that they reflect reality. Given the historical track record, I would recommend making plans based on cohort life expectancy, which tends to be about seven years longer than the period life expectancy statistics most of us have previously encountered. Having a longer time horizon changes the cost-benefit analysis for many kinds of investments. People who can reasonably expect to live longer have stronger incentives to invest in education and skill-development (of all kinds), since they have more time to reap the rewards. Moreover, longer-lived people also have good reasons to save more to enjoy their presumably longer retirement, and to make advance legal arrangements for their care in their last years. Additionally, people who expect a longer life might find it worthwhile to tolerate longer job searches and more career switches in order to attain an ideal-fitting job, even if the cost is longer periods of frictional unemployment between jobs and perhaps more mid-career education. Similarly, living longer might even increase the incentives for monogamous partner-seekers to search more cautiously and for longer, even at the cost of short-term loneliness, in order to increase the chances of finding a partner who they are compatible with in the long run. Beyond this, the expectation of a longer life disincentivizes certain forms of dangerous risk taking, since you perceive yourself as having more future life to lose. I am sure there are other important ways in which embracing a longer time horizon might nudge us to adjust our life plans as well. With a longer life, the payoff to patience is higher. Adjust your plans accordingly.
References:
Morgan, Edward. 2019. UK Office for National Statistics. Period and Cohort Life Expectancy Explained. https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/lifeexpectancies/methodologies/periodandcohortlifeexpectancyexplained (April 3, 2022).
You'll probably live longer than you expect
There's also a substantial difference between life expectancy at birth, and life expectancy at age 25 (or 40, or 60, or 80).