Skip to content

Will the pandemic cause a decline in births? We’ll be able to resolve this particular debate in about 9 months . . .

The fallacy of the one-sided bet

I’m gonna be talking about a news article and research paper asking the question, “Will coronavirus cause a baby boom, or is that just a myth?”

And my problem is the fallacy of the one-sided bet: By asking the question, is there a positive effect or is it zero, you’re not even considering the possibility that the effect might be negative.

The story

Joseph Delaney writes:

Early on in the pandemic, one of the worst takes is that the “stay at home” orders would result in a baby boom. This was a terrible take. Let us consider the countervailing forces:

1. Harder to meet partners, so this “baby boom” is mostly among people already living together

2. Those living together are under tremendous stress, which is not good for marriage. Obviously, if you are considering a divorce that is poor timing for children

3. There is suddenly a ton of financial uncertainty.

4. We got to see childcare collapse and schools turn into an unreliable form of childcare. If you are having trouble with two jobs at the same time due to unreliable (or no) childcare, that seems like a bad time to add children to the mix

5. Who wants a ton of medical center encounters during an actual pandemic?

6. In the United States, health care is linked to employment (the thing that the economic uncertainty also influences) and who wants to take risks with that?

The actual baby boom occured after the period of uncertainty and suffering was over (World War 2) and in the context of an economic boom. Sure, after years of deprivation, the sudden appearance of opportunity is going to make many people consider children. But going into conditions worsening, birth rates were already at record lows.

Now I am not saying that it is necessarily bad to have a falling birth rate. There are a lot of complex issues that go into whether or dropping birth rates are a net positive. But this was definitely one of the predictions that seemed to look at very surface effects and make a very naive extropolation.

Delaney links to a news article by Marie Puente that’s pretty good. First is the title, “Will coronavirus cause a baby boom, or is that just a myth? Prepare for jokes, if not babies!”, which has the appropriate tone of skepticism. Then in the article, Puente writes:

The stay-at-home coronavirus orders are unprecedented in America but the speculation about what may ensue nine months hence is not: Prepare for a baby boom!

At the least, prepare for jokes about a baby boom.

Or maybe not. Perhaps it’s all an urban myth propagated (no pun intended) mostly by baby boomers, that giant cohort of Americans born in the years after World War II when everyone came home and got busy again.

Nevertheless, it’s fun for people to speculate about a baby boom in the midst of the coronavirus pandemic. After all, what else do they have to do, stuck at home and bored?

Speculate, that is. Not the other thing. Because think about it: Even a mild version of coronavirus could make one sick enough to take to bed, and not with baby-making in mind. . . .

Still, there’s little data to support the claim that catastrophes – regional power blackouts, hurricanes and snowstorms, terrorist attacks or global pandemics that force people to stay at home for extended periods – lead inexorably, as night follows day, to a quantifiable jump in births months later.

But after every disaster, including this one, people think there might be such a boom so naturally the media rush to feed the speculation. Just as quickly, sober scientists carefully play down the hype. . . .

The article then cites several experts:

“On the whole, it’s unlikely that America will see a coronavirus baby boom – but we could see a baby blip,” wrote Richard Evans, associate director in the graduate computational social science program at the University of Chicago, in an op-ed column in The Washington Post last month. “Given all the previous evidence on how different types of catastrophes affect our fertility, it seems likely that we can expect a small increase in births as a result of the coronavirus.”

How small? Nationwide, he thinks there could be a 2% increase . . .

A coronavirus baby boom is “very unlikely,” agrees Philip Cohen, a professor of sociology at the University of Marylandwho specializes in family structure, marriage and divorce. Even if people cooped up had sex more often, he says, frequency isn’t what counts – it’s contraception.

“Lots of babies are born to couples not living together, who presumably (because of social distancing) are less likely to have sex and children now,” Cohen says. “So even if a few people accidentally or on purpose decide to have a baby now, they will probably be outnumbered by the lost births from people meeting less, having sex with non-residential partners less and deciding now is not a good time.”

Sheeva Talebian, a fertility/OB-GYN specialist and fertility expert for Women’s Health magazine, says she thinks people will hesitate to get pregnant right now due to stress, fear of infection or financial worries.

“Yes, there have been baby booms during times of enforced togetherness at home, but on the other hand, people tend to postpone kids when they are insecure about the future,” adds Stephanie Coontz, an expert in family studies and a professor emeritus at the Evergreen State College in Olympia, Washington.

“Birth rates generally fall during recessions and depression, and since this pandemic is causing serious and likely long-lasting economic hardship, I don’t expect many people to try for a child,” Coontz says. . . .

Even if there are more pregnancies, there may not be more babies, says Sarita Bennett, president of the Midwife Alliance of North America, a licensed family practitioner and certified professional midwife who is trained to do home births.

“Statistics that show when there is panic or terror, the very normal mammalian response is an increase in preterm births and miscarriages,” Bennett says. “The more you feed the panic, the more of a (negative) effect on pregnant people.” . . .

“I think it is going to be 50-50” more births and more divorces, says Randall Kessler, an Atlanta matrimonial attorney. “We are seeing it. We are fielding at least twice as many calls per day as in normal times for people who are considering a divorce.

“And this is historically consistent. In times of crisis, or when tensions are increased, divorces always seem to rise. It has already occurred in China. In the United States, we learned lessons from disasters like Hurricane Katrina and the 2008 recession.”

It’s interesting, though. The news article interviews 6 experts, of whom 3 think there will be fewer babies, 2 expect to see about the same, and 1 who expects a slight increase. But (a) the article leads with the one expert whose view is different from all the others, and then (b) presents the second expert as in agreement with the first, even though they are not in agreement! So, the article is good, but it seems to me that it’s slanted toward a particular story—there will be a baby blip, but it will be a small blip—even though the expert consensus seems to be that the number of births will actually decline, not rise. Indeed, looking carefully, it seems to me that the scholars cited in the article who are the most experts in the area are Philip Cohen and Stephanie Coontz, both of whom expect a decline in births.

So the real story should not be: There might be a little baby boom, but just a blip. Rather, is should be: We can expect a reduction in the number of births.

Clicking through

I was wondering about one claim in the article, “Given all the previous evidence on how different types of catastrophes affect our fertility, it seems likely that we can expect a small increase in births as a result of the coronavirus,” given that it ran counter to the statements from the other experts.

The link is to a Washington Post article by computational social scientist Richard Evans, which says:

In 2010, I [Evans] published a study with co-authors Yingyao Hu and Zhong Zhao that tried to measure whether baby booms result from hurricanes along the Atlantic and Gulf coasts of the United States. We had read of anecdotal evidence suggesting births increased in Florida after the devastating 2004 storm season. And we thought this coastal region was the perfect laboratory for examining people’s proclivity for procreation. We had data on many storms over a multi-year period. We studied a diverse set of mothers, from several income levels, education levels, races and marital statuses. This area also experienced storms at different levels of severity — tropical storm watches, tropical storm warnings, hurricane watches and hurricane warnings — lasting for different durations. In short, it was an ideal data set.

Our research yielded two different, but intuitive, results. For low-level storm advisories such as tropical storm watches, we measured a positive and statistically significant effect on births nine months later. An extra 24 hours of tropical storm watch advisory resulted in a 2 percent increase in births. This is evidence of the old New York City blackout hypothesis: When the lights go out, stores are closed and TVs are off, we make more babies.

On the flip side, we also found that an extra 24 hours of the most severe advisory — hurricane warning — resulted in a 2 percent decline in births. You can’t make babies if you’re running for your life. . . .

Given all the previous evidence on how different types of catastrophes affect our fertility, it seems likely that we can expect a small increase in births as a result of the coronavirus. For now, most people are home, not suffering from covid-19. We are more focused on community and the preservation of ourselves and others. This is analogous to the low-level storm advisory effect, or the situation examined in the Oklahoma City study. Of course, those infected with the disease are experiencing a more severe catastrophe, analogous to a hurricane warning — so we would expect those couples’ fertility to be lower.

This description made me suspicious: a positive effect for low values of the treatment but a negative effect for high values? It’s not that this makes no sense; rather, my problem is that such nonmonotonic effects are really hard to estimate and are very sensitive to where you draw the cutoff. Here’s an example (a claim about sports performance) from a few years ago, and there was another about the deterrent effect of the death penalty that had a similar problem with an overfitted nonmonotonic model.

I clicked through to take a look at the cited article, “The fertility effect of catastrophe: U.S. hurricane births,” and I am concerned about the many arbitrary choices in the model that could have large effects on the inferences. This is particularly a concern when results are reported in terms of statistical significance. In the absence of theory, what you’re getting is tons of noise; in the presence of theory, you’re getting the ability to confirm the theory via forking paths. Either way, I’m skeptical that such results would generalize, either to future hurricanes or to other catastrophes such as pandemics. Also the article has no scatterplots, so it’s hard for me to get a sense of what the data are saying.

In summary

I’m inclined to agree with the two experts quoted who thought that the pandemic would result in a decline in the birth rate, at least in the short term.

But I don’t really know.

We’ll learn more in 9 months.

Actually we’ll learn more right now, given that I’m posting this on blog delay and it’s not scheduled to appear until Feb 2021.

Leave a Reply