What Are the Consequences of Low Fertility?

I had a twitter exchange last week with Noah Smith and Adam Ozimek about fertility. Basically, Noah suggested that US fertility was only a little bit below replacement, likely to rise, and that, anyways, being a bit below replacement rate fertility doesn’t have big population consequences. I suggested that literally none of these suppositions were true: U.S. fertility is presently quite low, it is not likely to rise naturally, and in fact is already low enough that population consequences are severe. Adam challenged me to model it out.

Let’s start with the “what is US fertility today?” question. Here’s TFR:

The red lines are probably most accurate; the light green line is woefully wrong as CPS doesn’t ask about birth, I just impute from “own children age 0 living in parent’s household,” so it misses kids living outside their parents’ household or some quirks involving age definitions.

U.S. total fertility is not at record lows, though total fertility in parent households may be; i.e. more and more newborns end up somewhere other than their parent’s household. However, total fertility has fallen dramatically since its recent peak in 2007, and that decline has continued even as the recovery has picked up speed. We can get monthly birth data, and monthly estimates of the reproductive-age female population, through June of 2017, so we can calculate the General Fertility rate, which does not exactly follow TFR, but usually doesn’t move in the total opposite direction, and controls for at least some population aging effects.

And lo and behold, GFR has fallen meaningfully in 2017.

So TFR has fallen, and will almost certainly continue to fall in 2017. And given that most economists expect a recession in the next few years and recessions have negative fertility effects, any recovery in fertility in 2018 or 2019 will probably be short-lived, if it occurs at all. Ergo, we cannot expect any durable fertility boom until at least 2020 or 2021. And even then, a fertility recovery remains speculative.

Any responsible demographic forecast must assume fertility falls in 2017 and at best plateaus until 2019 or 2020; further decline is highly probable as youth fertility still has space for further declines and trends driving diminished fertility, like increased use of IUDs, delayed marriage and childbearing, and increasing female educational attainment, are likely to continue. Desired fertility continues to inch lower as well.

So: fertility is low, and the most responsible expectation is that it will remain low until the mid 2020s at least.

However, total fertility can be impacted by timing: women’s reproductive years may include both high and low fertility periods. As a result, completed fertility is not synonymous with total fertility. However, I have shown before that total fertility in primary reproductive years weighted by share of births in each year across the whole population is a reasonably good predictor of completed fertility. That is, if we know a woman’s age cohort, and we know the age-specific fertility rates across her whole lifespan, then we can predict her completed fertility to within a reasonable degree of precision, basically we’re just dealing with normal stochastic errors at that point.

We do not know that information in hindsight because I don’t have exact age-specific birth rates or exact cohort-specific female populations going back the last 60 years or whatever. But I do have that data for the last ~10 years, and I can invent that data for the next 40 years. And as such, I can create a plausible forecast of completed fertility using my already-demonstrated population forecast model for women who reach age 15 any time between 2005 and 2025.

But now we are into forecasting. So let’s talk about the effect of different TFR forecasts.

Impacts of Different Fertility Rates

I’m going to plug in 5 different scenarios, shown below.

For the record, I find somewhere between the gray and the red lines to be most plausible.

What impact would these have on population forecasts? Well, if I change my fertility parameter but change no other parameters (meaning, I don’t make offsetting adjustments to counteract the fact that my model includes an interaction effect with migration: lower fertility induces slightly higher net immigration ~20 years later in my model), the following population estimates are what I get:

As you can see, the “big fall” scenario, which would have the US experiencing similar fertility trends as Europe and the former Soviet Union have experienced the last few decades, results in population decline setting in by the mid 2030s. The next worst scenario, the “Static Low” scenario, has fertility declining, then simply staying stagnant at low, levels, and has decline setting in a few years later. The “Mid Fall” scenario has TFR bottoming out around 2025, then posting robust growth, returning to replacement-rate by the 2050s. This scenario has a greatly diminished growth rate of population, but does manage to eke out some growth. Then we get our other scenarios, showing good growth. But crucially, even if we reach well-above-replacement-rate fertility, population growth rates will decline! This may surprise some readers, but the point is that to maintain growth rate X for the whole population while a larger share of that population is outside of reproductive years (retirees) means fertility for reproductive-age people must be even higher. In other words, maintaining historic rates of labor force growth will require progressively higher and higher rates of fertility, or higher and higher rates of immigration. At least until the baby boom cohort has died out of the population pyramid.

To me, these are large effects of comparatively modest changes in fertility. Entirely plausible fertility scenarios end up yielding very pessimistic population forecasts; even the most optimistic fertility scenario yields a pretty sharp bend in the population curve. And, crucially, again, my immigration forecasts are probably a bit on the optimistic side: I have annual net inflows remaining around a million people a year in every scenario and year. That very well may not happen as more countries develop out of peak migrant-sending income/fertility bands, and as the new countries developing into peak-sending territory are much more distant and easy to control with policy rules (Africa, Asia).

And what happens to completed fertility under each scenario?

Under all but the big fall and static fall, it eventually rises to reach or exceed current levels. For the record, this chart reflects birth cohort-specific completed fertility, so “2005” here means women born in 2005, so having babies 2020–2055. My forecast stops at 2060, so at 2060 I just flatline age-specific fertility straight out. So for women born after 2025, you’re not seeing a real forecast, but just a convergence to TFR.

But look at the years we can forecast best, completed fertility for women born, say, 1985–2005, since we already have or can reasonably expect to forecast the age-specific fertility rates for at least a few of their high-fertility years. Even under the highest-fertility scenario, completed fertility drops sharply in early years. Under virtually any plausible scenario, complete fertility for these women is at best equivalent to the 1950–1970 cohort, and probably well below the 1970–1985 cohort women.

In other words, under any plausible forecast, today’s childbearing women will end up having a lot fewer kids than previous women, even women in the recent past. This is true even though I forecast very robust gains in late-in-life childbearing.

Conclusion

Even a modest decline in fertility results in literally tens of millions of fewer people than otherwise, meaning a seismic impact on how many cities can expect to forecast growth, the distribution of political power, and the rate of GDP growth. And even with a big increase in fertility, growth is going to slow down markedly. Under virtually any scenario, women born 1986–2005 can be forecast to have substantially diminished completed fertility, although there do exist non-far-fetched scenarios where completed fertility recovers to current or higher levels.

To me, this means I won: modest fertility changes have macro-significant impacts.

Check out my Podcast about the history of American migration.

If you like this post and want to see more research like it, I’d love for you to share it on Twitter or Facebook. Or, just as valuable for me, you can click the recommend button at the bottom of the page. Thanks!

Follow me on Twitter to keep up with what I’m writing and reading. Follow my Medium Collection at In a State of Migration if you want updates when I write new posts. And if you’re writing about migration too, feel free to submit a post to the collection!

I’m a native of Wilmore, Kentucky, a graduate of Transylvania University, and also the George Washington University’s Elliott School. My real job is as an economist at USDA’s Foreign Agricultural Service, where I analyze and forecast cotton market conditions. I’m married to a kickass Kentucky woman named Ruth.

DISCLAIMER: My posts are not endorsed by and do not in any way represent the opinions of the United States government or any branch, department, agency, or division of it. My writing represents exclusively my own opinions. I did not receive any financial support or remuneration from any party for this research.