How can fertility be down even as the number of women who are having children is going up? There are complex statistical reasons for this, but the main cause of this confusing discrepancy is the age at which women are having children. Women are having children later in life — the median age for having a first baby is 26 now, up from 23 in 1994 — and this delay causes annual birth rates to go down, even as the cumulative number of babies per woman has risen…
Another factor, Livingston said, is the drop in teen birth rates, with black women seeing the biggest drop in that category.
See the Pew report here. An additional part of the explanation is that there are multiple measures at play here. A Pew report from earlier in 2018 explains:
But aside from this debate, the question remains: Is this really a record low? The short answer is: It’s complicated.
That’s because there are different ways to measure fertility. Three of the most commonly used indicators of fertility are the general fertility rate (GFR); completed fertility; and the total fertility rate (TFR). All three reflect fertility behavior in slightly different ways – respectively, in terms of the annual rate at which women are presently having kids; the number of kids they ultimately have; or the hypothetical number they would likely have based on present fertility patterns.
None of these indicators is “right” or “wrong,” but each tells a different story about when fertility bottomed out.
Measurement matters and the different measures can fit different social and political views.
I wonder if part of the issue is also that there is a clear drop in births from the earlier era – roughly 1950 to 1970 which we often associate with Baby Boomers – but the last 3+ decades have been relatively flat. This plateau of recent decades means researchers and commentators may be more prone to jump on small changes in the data. Many people would love to predict the next big significant rise or fall in numbers but a significant change may not be there, particularly when looking at multiple measures.