The Regularity of Demographic Transitions

[This is part of my series on Thomas Malthus’ “Essay on the Principle of Population,” first published in 1798. You can find an overview of all my posts here that I will keep updated: “Synopsis: What’s Wrong with the Malthusian Argument?”]

Demographic transitions totally stomp Malthusians. Actually, they should not exist at all. But they do. Not sometimes and somewhere, but practically all the time and everywhere. Not only over the past few hundred years, but throughout human history for 100,000 years or more.

There were predictions that Muslim countries would not have demographic transitions because of supposedly “Muslim fertility.” It’s in the religion, you know. Never mind that Muslim countries had practically no population growth between the years 1000 and 1500. Some even shrank. However, also Muslim countries have had their own demographic transitions lately. So the prediction has moved on. Now it is that African countries might not have demographic transitions. But I am quite confident they will have them, too. And then you can forget about all the r/K-selection theory nonsense.

— — —

France was the first country in Europe that experienced a demographic transition in modern times. It began around 1800 and was already well advanced by the middle of the 19th century. Other European countries were still growing. For example, Germany surpassed France later in the century, at the start it had been less populous. The French worried about that.

Some German nationalists in the 1870s, though, were gloating because they thought France was doomed. But population dynamics have a good sense of humor: Right around that time also Germany began its own demographic transition that caught up with developments in France already before World War I. By the 1920s, the Germans were in full panic mode because now Germany was doomed. But last I looked, both Germany and France were still there and doing fine.

Another variant of this fearmongering is to point out how the shares for continents change. Europe reached a peak around 1900 with about a quarter of world population. Since then it has gone down to about 10%. The share for Asia has gone up, and especially that for Africa. Personally, I am totally not fazed by this. It’s great that there are more people in Africa and in Asia, which shows they are doing better. But apart from the judgment, also the whole take is wrong as I will now demonstrate.

Here are the shares for continents since 1600 (underlying data here). As there was a lot of migration from Europe to the Americas and also Oceania, it is hard to disentangle the developments for these continents. I just add them together as “Europe & Americas” (which includes Oceania, that was only too awkward in the graph, no disrespect). I have also applied some smoothing because there were fewer data points before 1900, but the exact figures don’t matter for my conclusions:

As the graph shows, Europe may now have a low share of world population. But a good part of the explanation is that much of the action shifted to the Americas and Oceania. Actually, the share for those continents together kept rising until the first half of the 20th century. This relative increase already started around 1700.

Asia also grew its share, even before Europe, the Americas, and Oceania did. But as the latter picked up speed, Asia fell back relatively until about 1950. At the same time, Africa saw its share decrease with a minimum around 1900. Actually, its population stagnated from 1600 until the middle of the 19th century while the rest of the world grew. Tell me more about r/K-selection theory. Only lately has there been a change. Africa and Asia expanded relatively, but that is already over for Asia.

And here is the remarkable thing: Look at the percentages in 1600 and the percentages now or where they are going to. Those are practically the same shares of total world population! So after 500 years, it looks like all this will have been a wash. The relative shares are back to where they once were.

There is only one explanation what this means: All three parts of the world will have grown practically by the same factor over half a millennium. There were vast differences in between, but after all is said and done, it mostly cancels out.

That is not only so on such a high level of aggregation. If you look at demographic transitions for individual countries and include emigrants and their descendants, they all grow by a factor of around 12 over a whole demographic transition lasting a few centuries (here from 1600 until 2000). Some grow less, some more. You find a range here from perhaps 5 for France, 7 for Japan to around 18 for Sweden and 19 for Egypt. But when you break that down to annual growth rates or average fertility, all countries are extremely close together. I will develop this more in detail in another post.

To stress the main point here: A population that goes from preindustrial conditions to those in modern industrial societies grows by a factor of about 12. There is a range, but it is much to narrow that there could be a lot of deviation from the general pattern. Demographic transitions are remarkably similar across countries, cultures, religions, etc.

While the outcome is almost the same, the exact development can vary a lot. Here are annual growth rates for the three parts of the world (average rate between data points, shown at the end, and again with smoothing):

As already noted, the population of Africa hardly grew at all until 1850. The pace then picked up slowly. But it took until 1900 to match Asia, and until 1950 Europe, the Americas and Oceania. Since then there has been fast growth, but with a slight decrease lately.

The development for Asia was different. It had moderate growth already early. Then there was an sharp increase from about 1950 on, but growth rates are by now in fast decline. The development in Europe, the Americas, and Oceania was much smoother. That part of the world started out already with comparably high growth, though for a long time only slightly ahead of Asia. The peak was then lower, and the decline started earlier on. Actually, the three peaks come in a row with a few decades between them.

But the overall increase over the whole 500 years is very similar as attested by the respective shares of world population. So while it is easy to be distracted by the steep increases and the height of the peaks, it does not matter much over the longer run because the eventual outcome is practically the same. Europe, the Americas, and Oceania had the same demographic transition as also Asia and Africa have or are having, it was just more drawn out and came earlier.

Actually, if you think about it, that makes sense. The Industrial Revolution was pretty slow compared to what happened in the 20th century. But Europe and then the Americas and Oceania had the effect first. When modern development came to Asia, which had its own history of development, and especially to Africa in the 20th century, it was explosive because you could have what had taken longer before in a much shorter period of time.

Here is also another bitter pill for the r/K-fans. If there are fundamental differences for the people across the world that lead to more or less fertility, where do you measure that? If you look at a cross-section now, it seems like fertility ranks: Africa > Asia > Europe/Americas/Oceania. But if you take the cross-section in 1850, it is exactly the other way around: Europe/Americas/Oceania > Asia > Africa. What is it then?

I would say: neither. You have the same development that only works out differently over time, comes earlier or later, and may be more drawn out or more compressed. What you see in the cross-sections is just an artefact and not a constant fact about the people in the respective parts of the world. They actually behave practically in the same way as you can conclude from the eventual outcome. After half a millennium, it will have been practically a wash.

It is almost like we are all humans!

— — —

The conclusion that I will explore in further posts is this: If demographic transitions are variable in their development, but always lead to practically the same result, then this cannot be so because of how they work out. That only shifts things around or drags them out or compresses them. The outcome must depend on the move from conditions in a preindustrial to those in a modern industrial society. When a population goes through this development, however it proceeds in detail, it just grows by a certain factor.

There is a range because already small deviations make a difference, but as I will demonstrate in another post, those have to be really small. To give you the flavor of the results: If I normalize average fertility from 1600 to 2000 to a replacement level of 2.1 as nowadays, the range for individual countries goes from 2.3 to 2.7. That is less than half a child per generation between the extremes, and most countries fall in a range from 2.5 to 2.6. Good work, humankind, that’s very precise.