How Long Until We’re All Amish?

Never Gonna Happen, But It’s a Fascinating Question!

The Amish are a fascinating group for a demographer. Living a largely pre-modern-type live surrounded by modern society, and modern record-keeping, many demographers treat the Amish as a kind of “living fossil,” that is, a look at “natural fertility conditions,” that is, fertility conditions without birth control or industrialization. I think that view is mostly wrong, but explaining why is a bit tricky, and getting data on these folks ain’t always easy.

So let’s look at the Amish. First of all, how many are there?

There are a lot of studies out there that estimate the number of Amish people. Some agree, some disagree, but the differences aren’t huge. I will basically take the estimates of the Amish Studies Center at Elizabethtown College in Pennsylvania, supplemented by a few other sources on earlier periods, and directly interpolated for missing years.

And that folks is an exponential growth curve. You don’t see one of them suckers in the demographic wild every day, and, when you do, it’s usually some kind of new-settlement of virgin country or something… or a developing country with few mass-catastrophic events.

Here’s population growth (spiky because of my linear interpolations):

Growth rates have risen and fallen, but there’s no strong time trend here. You may also be wondering, “With growth rates that high, how long until Amish population doubles?”

Weeeeelll…

So a reasonable presumption would be that the Amish population doubles every 15 to 30 years. In other words, every generation is about twice as big as the one before it. This means the average woman must be having at least 4.2 kids assuming no conversions. But if you then assume, as most literature suggests, that about a quarter of Amish youths ultimately leave the church, and that conversions into the Amish lifestyle are very rare, then you arrive at a situation where long-run fertility of the Amish in the 20th century must have been 5.5–6.5 children per woman.

But do we have any data to pack that up?

Heck yes we do!

One team of researchers uses Census data to identify women who had completed their fertility years in the 1960, 80, and 90 Censuses and who spoke Pennsylvania Dutch, the distinctive language spoken almost exclusively spoken by Amish people. As a second specification, they narrowed this sample to women who had no phone in their home in the 1980 and 90 censuses. Here are their resulting estimates of completed fertility by birth cohort for Old Order Amish (a subset of the most “plain” Amish), vs some key baselines. I have also included completed fertility rates directly measured in the Amish communities of Geauga, Ohio, by a different team of researchers. That second team uses Amish community records of births to estimate historic birth trends.

The most reliable data, the Geauga data, shows completed fertility of over 7 kids per woman for those born in the first half of the 20th century, then a decline. The Census-based data is lower, but still confirms very high fertility, and a decline in the postwar-born generation. Interestingly, simply speaking Pennsylvania Dutch is not a powerful indicator of super-high fertility. It’s Pennsylvania Dutch and lack of a phone that seems to be the really strong indicator of Amish-type fertility.

So with that in mind, I did a little digging myself in IPUMS. What are current fertility rates for Pennsylvania Dutch-speaking-women-with-no-phones?

The above chart doesn’t show any indicator for Amish women. Instead, it shows some food-for-thought. Rural women already have higher birth rates than women generally, and the Amish are overwhelmingly rural.

Another interesting comparison is Hispanic women. Hispanics are classically considered to have higher fertility rates than non-Hispanic women, so it should be interesting to compare them to the Amish.

Now then. Let’s add the Amish. We’ll do one line for Pennsylvania Dutch-speakers, and one line for Pennsylvania Dutch-speakers who also have no phone.

aaaaand there you have it folks. Fertility is still extremely high among Pennsylvania Dutch speakers, and especially high among those with no phone. We’re talking average TFRs from 2001–2015 of about 6 or 7 births per woman. That’s genuinely remarkable, and that doesn’t include twinning. That’s just delivery count.

Now here’s a cool trick. The Geauga Amish study was able to give us age-specific fertility rates by birth cohort, which we can reallign into period-specific fertility rates, to get estimated Amish TFRs for a limited number of years. Here’s the result:

But those ACS-derived TFRs are so bouncy. Let’s smooth them out by averaging across several years.

Now, one last thing. Let’s, just for fun, do a few more adjustments. Let’s assume actual Amish fertility in recent years is the average of our 2 estimators. And let’s assume it’s a contiguous series with the Geauga series. And let’s see if changes in that series mimic changes in U.S. fertility generally.

Well, well, well. What have we here.

At least in the Geauga, Ohio Amish settlements, the decline in fertility followed national fertility trends very closely. Here’s a fun fact: the Amish don’t use most forms of birth control or abortion.

Now, this doesn’t mean Amish fertility fell as low as U.S. general fertility; it simply means that Amish fertility fell as much as U.S. general fertility.

In the modern period, we can see that, from the early 1980s to 2000, Amish fertility had actually fallen way more than U.S. TFR on the whole. Then it spiked in the late 2000s, and has fallen since.

But let’s zoom in on the 2000s for a second and index both Amish fertility and U.S. TFR to 2001.

You seein’ what I’m seein’? Cuz what I’m seein’ is that Amish fertility is pretty well correlated with U.S. TFR on the whole. And no, the Amish are not a big enough population to drive this trend: removing the Amish from the sample doesn’t change U.S. TFR essentially at all.

So this is actually very important. See, the Amish are cut off from certain parts of U.S. society (many cultural norms, certain technologies, etc), but they aren’t cut off from others. The Amish have farms and businesses that have to make money, and most Amish consider it morally acceptable to manage the spacing (and to a limited extent even the total number) of childbirths. So the Amish are impacted by the business cycle.

I have been strongly on the side saying that the decline in fertility in the U.S. is driven by technological changes (LARCs and emergency contraception) and by cultural changes; that the business cycle is not the primary cause anymore.

But the evidence from the Amish suggests otherwise. Even a group that had contraceptive use and cultural norms held roughly constant experienced a similar change in fertility as the U.S. general population. The only thing that they would have been exposed to is economically-transmitted fertility shocks via changes in economic returns from their businesses which sell to non-Amish people. This biases in favor of economic cyclicality driving fertility trends, and against culture.