More Thoughts on Falling Fertility
And Some Responses to Critiques
My “Baby Bust” post got lots of attention; way more than I expected. It has thus gotten lots of follow-up questions and critiques. There are four main questions I’d like to address:
- But Lyman, why should we even care about slow/negative population growth?
- But Lyman, isn’t this just a timing effect, and completed fertility will bounce back?
- But Lyman, can’t we just admit tons of immigrants?
- But Lyman, shouldn’t you listen to women instead of just demanding they have more babies?
Is Overpopulation a Problem?
Not in the United States
I have two articles outstanding at other publications that will elaborate on this question in great detail. But I will offer here an extremely succinct summary of the overpopulation question.
Globally, overpopulation creates sustainability risks. In the United States, it does not. We have superabundant land and natural resources and an economy that is improving its carbon efficiency at breakneck pace. Indeed, by the end of the century, given plausible carbon efficiency and growth parameters, relocating a person from a middle-income country to the United States could be carbon-reducing on average. More broadly, there is no plausible population scenario that can reduce carbon emissions to a scale necessary to prevent catastrophic warming, partly because the effects are too long-delayed given population momentum. Even modeling carbon efficiency, growth limiting, and population changes together reveals that population limiting barely adds anything to efforts to prevent climate change. I will demonstrate this empirically in a future post, but the basic story is simple: to prevent climate change, we must dramatically reduce fossil fuel consumption. That is the only solution.
More importantly, we should want higher fertility in the United States because:
- Women want it (elaborated below)
- Society needs it (for long term fiscal obligations)
- The economy demands it (to maintain growth rates sufficient to enable normal amounts of economic dynamism)
- Global security depends on it (because odds are we have to fight a mass-casualty war within the next 2 centuries)
I’m trying to keep this post brief, and I’ll explain these further in other posts elsewhere, so that’s all for now. Sort of a teaser. But I’ll tell you, the long-run cost of a 0.1 decline in TFR today is somewhere between $30 and $130 billion per year in social security taxes per year by mid-century. That means benefit cuts or tax hikes.
Some readers suggested higher fertility would make housing prices rise. That’s bogus. Housing prices are high because of restrictions on land use, not babies: tellingly, we’ve seen sharp rises in housing prices during a period of rapidly falling population growth rates! There is fundamentally no reason to think national population growth has a meaningful impact on housing costs that can be discerned apart from land use rules.
Will Millennials Have Fewer Children?
Leslie Root, a PhD student in demography and a Russianist, disagreed with my take on fertility. Her point is simple: she says that all I’m observing is a “tempo effect.” That is, TFR projects current fertility rates for each age group into the future, with age-specific birth rates held constant. TFR, then, is a kind of “naive forecast” of completed fertility.
But of course, age-specific birth rates are not constant over time. So what happens if they change over time? What will happen to completed fertility?
I’ve discussed this extensively before. Here’s one where I show much of diminished completed fertility is due to changes in marital status: changes that are accelerating and becoming more extensive with time. Here’s one where I impute historic completed fertility from well before the current period, showing the very long-run decline. And here’s one where I simulate the effect of different future TFR trajectories on population and completed fertility.
Let me do another round of simulations. Leslie responds to my dismal scenario by suggesting that the impacted women were just delaying having kids. It’s true: we are seeing rising fertility among older moms! We can calculate age-specific birth rates from two major sources: the American Community Survey and the CDC’s births data. Here’s a comparison of their data from 2001–2017. Notably, ACS does not track births among females under age 15 or over age 50, while CDC does. These births are an exceedingly small fraction of total births.
As you can see, there are marked differences, with ACS tending to have lower estimates of younger fertility, higher estimates of older fertility. Part of this is due to sampling frame: CDC reports age of the mother at birth, while ACS asks women if they’ve had a birth in the past year, so women could be up to a year older when ACS measures them than they were when the child was born. There are a number of other survey biases at work too, like the ACS not updating for more recent underlying population estimates.
But on the whole, while levels vary, trends look very similar! Falling fertility for younger moms, rising for older!
So then, what if these trends continue? From here on out, I’ll be using just the CDC age-specific fertility rates (ASFRs). I will run three ASFR scenarios, shown below. Here are their total fertility rate outputs:
Scenario 1 makes a very simple assumption: current age-specific trends continue. Basically, the average annualized percentage decline/increase in age-specific fertility over the last 5 years continues. So young births keep falling, older ones keep rising. I taper off growth near the middle of the century to avoid preposterously high or low ASFRs. Scenario 2 assumes declines continue for younger groups, but assumes we get even more aggressive increases in older fertility than we’ve had in recent years. Scenario 3 builds on Scenario 2, but assumes that 20-something fertility stops its decline earlier than Scenario 1.
Put another way, Scenario 1 is “current trends,” Scenario 2 is “improved reproductive tech,” Scenario 3 is “Tech + Slightly Younger Marriages.”
Here are the ASFRS:
Now then. Let’s simulate completed fertility for each birth cohort from 1988 to 2020. Basically, we will have a woman be “born” in 1988, begin having a chance of having kids at age 11 (I’ll backcast under age-15 fertility from for birth cohorts 1988–1992), and continue having a chance of having kids until age 55. In each year, her odds of having a kid are the ASFR for her age.
Using these ASFRs, we can develop simulated CFRs by birth year:
If current trends continue, a woman born in my birth-cohort, 1991, so “solid Millennial” territory, will have 1.96 kids. So that’s not that far below replacement! Woohoo!
Where it gets dire is the generation just a bit younger than me. If current ASFR trends continue, the Kids These Days will end up having well under 1.75 children per woman. That’s low. That’s comparing to Russian completed fertility at it’s lowest ebb, as Leslie showed in her article. And again, if you think the Russian experience was okay, then, well, you are definitely a minority opinion, and almost certainly wrong. The extent to which almost every low-fertility society implements pro-natal incentives suggest there is endogeneity in the political system: fertility falls, which makes people extremely unhappy, and they communicate it via the political system, demanding support to obtain their fertility goals. There are actually very few large low-fertility societies out there that don’t have generous incentives or campaigns to boost fertility in place. Now, most of these are of minimal effectiveness! But their mere existence suggests that lowest-low fertility creates direct disutility for voters, at a minimum.
Here’s that data shown again, with historic completed fertility shown:
As you can see, completed fertility fell sharply after the 1930s birth cohort, because the baby boom ended. Then CFR was below replacement through the early 1970s birth cohort. Since then, it seems likely that CFR rose, and so the birth-cohorts for women born in the 1980s may even be above-replacement. But since then, CFR appears to have fallen, and is likely to continue falling. By the time we get to mid-1990s birth cohorts, we can expect record-low completed fertility. By the time we get to the 2000s birth cohorts, we’re talking post-Soviet collapse levels of fertility.
Maybe my ASFRs were too pessimistic! I’d love for folks who think this decline is 100% tempo to publish the ASFRs they think will yield near-replacement-rate completed fertility in the future, and explain how we’re going to get there. I would truly love to be convinced that my ASFRs are hopelessly pessimistic.
Final note: completed fertility has its own problems as an indicator. Some women die before reaching completion-age, meaning the fertility experience of women who reach their 50s is not necessarily the fertility experience of all women. Some women emigrate as well, either permanently or temporarily.
Plus, fertility experience may enter and exit the dataset non-neutrally. Imagine an imaginary kingdom that prohibits childbirth but gives $1 million awards for having a 3-year-old in your home: people would relocate to have a kid and return after having it. CDC-style data would show essentially zero fertility despite the population nonetheless having no shortage of 3-year-olds!
So: do U.S. women move abroad to have babies? Well… probably not? But maybe? That’s an analytically open question to me. I can see arguments either way!
Immigration can directly change fertility completion as well. Immigrants may have children in foreign countries then move to the U.S. Even though the births occurred elsewhere, and indeed the children may not be in the United States at all, they will show up in completed fertility statistics. Parents may also leave: people might have kids in the U.S., then emigrate! This is a huge problem in some localities like the Northern Marianas Islands where being born on U.S. soil confers citizenship!
I don’t know the extent of these problems in the aggregate. But the point is that while it’s true that current, ASFR-chained Total Fertility Rate is a tricky indicator, so is completed fertility! That’s why we need to use multiple indicators, as I pretty much always do. I didn’t use completed fertility for my 2017 post because we don’t have completed fertility data for people having kids in 2017.
Can’t We Admit Immigrants?
Maybe, But Not Really
Immigrants will not solve our fertility dilemma. Here’s total fertility rates in the U.S. according to the ACS for native- vs. foreign-born people:
Today’s immigrants are far more likely to be higher-skilled, and to come from lower-fertility countries in east Asia, than were migrants of even 15 years ago. Fertility rates are falling globally, so even Latin American immigrants today yield less of a fertility boom.
Now it’s true there’s still a gap, but the point is that immigration alone won’t solve the problem. It’s on a time delay vs. native fertility, but higher immigration really just kicks the can down the road by maybe a decade. Which, look, that’s awesome, and I’m a big fan of immigration! But this issue is gonna come and bite us eventually. So let’s kick the can a bit and in the meantime solve the underlying problem.
Plus, immigration is slowing down. Political factors, falling fertility in the rest of the world, increasing economic development in our traditional migrant-sending partners, and new-migrant-sending regions being more focused on Europe mean that the U.S. is likely to have underwhelming amounts of immigration in the future. This lever just won’t get you as much change as it used to.
To fix critically low fertility, you can’t only look to foreigners. You can look a bit to foreigners, for a little while, but at the end of the day you need long-term residents in America to have something nearer replacement rate fertility. Alternatively, the government could subsidize international adoption on a monumental scale and we could make countries opening themselves to such adoptions a key foreign policy goal that we demand in exchange for U.S. aid. That would be an interesting, if sort of creepy and exploitive, policy, but it could theoretically have a discernible impact on the demographic profile of the United States.
Women Say They Want More Kids
I got tons of responses saying that I should LISTEN TO WOMEN, and if fertility is low it’s low because THAT’S WHAT WOMEN WANT.
Well, I do listen to women. The National Survey of Family Growth shows women intend to have above-replacement-rate fertility, as Brad Wilcox shows:
Now, yes, the number of “intended” births is falling! But it remains waaaaay above total fertility.
But is intended births the right metric? “Intended” births includes a lot of other factors: husbands not wanting kids, insufficient finances, etc. “Intended” births are an absolute bare minimum estimate of how many kids women want. It is the hard-bottom revealed preference… and even then women do not tend to fulfill their intentions.
What if we give female survey respondents a bit more latitude, and ask them how many kids they’d ideally like to have? Well, here’s that:
But maybe there are age-specific factors there.
Thankfully, the GSS enables us to cut this data lots of ways! I’m going to show how many kids women actually end up having, grouped by how many they think would be ideal to have, by age, and by marital status. In other words, we can systematically use the General Social Survey to estimate how many women have fewer kids than they want, and how many have more.
That diagonal line is “attainment” fertility, meaning women have as many kids as they think is ideal. Above the line means women had more kids than they wanted. Below the line means they had fewer.
The first thing to note is that there’s a lot of colorful lines below the attainment line. This is because most women achieve less than their desired fertility.
However, it’s notable that there are some women on that left hand side: women of all ages and marital statuses who identify the optimal number of children as zero tend to have more kids than they want on average. Among women who think the idea number is 1, it’s more evenly split.
What’s also notable is that the lines do generally slope up: in other words, idealized fertility is a reasonably linear predictor of achieved fertility, especially for the married women who can most easily obtain their goals due to having a steady partner and household. Thus, while there are shortfalls, the very most basic demographic controls suggest that idealized fertility is a good indicator of womens’ actual, real preferences, not just an arbitrary number they give to survey-takers.
So then. How many women have more/less kids than they think ideal?
Doesn’t matter how you cut the data: below-ideal women are more numerous than above-ideal women. Now, this may be as it should be: the social and individual costs of above-desired fertility may be greater than below ideal, as above-ideal can lead to mistreatment, orphancy, ruined parental careers, and all sorts of problems. We don’t want to trick people into having kids they don’t want to have.
But, as we continue to help women avoid unintended or undesired births, here’s a suggestion: how about we facilitate them having desired births too.
I am listening to women when I write about fertility. Whether you look at intended or ideal fertility, both are above-replacement and far-exceed current fertility rates. Furthermore, this is a real, durable problem, not just tempo effects. Virtually any plausible scenario for fertility suggests that Millennial completed fertility will be low, and post-Millennial completion even lower. This will have negative social consequences on a huge range of outcomes and risk factors. I humbly submit we do something about it.
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.