Discussion about this post

User's avatar
Sarora's avatar

One Image: Cohort-Life Expectancy.

What ks going on by income quintiles across the cohorts is clear. Cohort-Life expectancies come closer to the actual "life years lived" experience than period-life expectancies. A period-life expectancy calculation basically imagines the current period's age distribution of mortality as a life itself to calculate the expectancy (under some very unreasonable assumptions). But people may go on to live longer or shorter than what the current period's age distribution of mortality may imply. The age distribution of mortality could be worse or better in the future.

But that poses a dilemma. If the bulk of the 1960 cohort is still alive, how does one know their cohort life expectancies accurately? You can't (without some assumptions). All you have is truncated information and a truncated life table. A cohort life table won't be complete until the entire cohort has perished. We are then left with the awfully adequate period-life expectancies.

One way to know how period-life expectancies (PLE) could mislead is to study the relationships between period-life expectancies in the past along with cohort-life expectancies (CLE) of the cohorts that have perished. Under what circumstances did PLE in -- say 1900 -- underestimate/overestimated how long a cohort ended up living in future years? Why did that happen? This has been done, though not with life tables, themselves, but with data on deaths/mortality to study aging.

Expand full comment
Philip Koop's avatar

"An open-and-shut case finally revealed the criminal behind the legend"

When the legend becomes criminal, print the crime.

Expand full comment
2 more comments...

No posts