The Total Fertility Rate Is Kind of a Nonsense Statistic

Periodically, the Census puts out something it calls synthetic work-life earnings (SWLE) estimates. To derive these estimates, what you do is you find the current median income at age 25, age 26, age 27, age 28, and so on until age 64. Then you add all of those median incomes together.

SWLE estimates are somewhat illuminating if your goal is to compare one demographic group to another. So for instance, you could find the SWLE of college graduates versus high school graduates or the SWLE of white people versus black people.

But SWLE estimates are not actually based on how much anybody or group of people earns over a lifetime. Nor is it an effort to project how much someone turning 25 this year will make in their next 40 years of life. It’s just sort of a fun spreadsheet exercise.

The total fertility rate (TFR) that people tend to cite in discussions of fertility is calculated the exact same way that SWLE estimates are. For any given year, you find the current birth rate of 15 year olds, 16 year olds, 17 year olds, 18 year olds, and so on all the way to 45 year olds (or sometimes 49 year olds). Then you add all of those birth rates together.

Contrary to popular belief, TFR does not tell you how many children women are having over the course of their lifetime. If you want to know that statistic, what you need to do is look at the most recent group of women who have exited childbearing years (e.g. 40 to 44 year olds) and see how many kids they had. This is called completed fertility and the Census has a fertility supplement with this information in it.

TFR might seem like a decent forward-looking proxy for completed fertility, but it suffers from some serious flaws.

For example, suppose you have a society with a high teenage birth rate. Through a variety of cultural changes and public policy tweaks, you manage to drive teenage births down way low and move those births into people’s 20s and 30s. In this scenario, completed fertility is unchanged as people are having just as many kids as they were before. But the TFR would initially decline because it would take some time for those delayed births to get picked up again in TFR.

This of course is not a hypothetical society, but is our own society. And it’s not just teenage births that have been pushed back. All births have been pushed back. Since TFR is initially unable to distinguish been fewer births and later births, TFR will systematically understate actual fertility until birth timing stabilizes again.

I am not saying anything people who study this don’t already know. But it’s not something people who write about this popularly seem to know. TFR is not completed fertility nor is it even a serious effort at projecting what completed fertility will be for the newest cohort of young women.

I understand why people don’t like to wait for completed fertility because that information necessarily comes out at a lag. But replacing a lagging stat with an inaccurate stat doesn’t help all that much and can cause more confusion than clarity.