In my three years forecasting with Metaculus, I have predicted on over one thousand questions. Many of these questions cover the same topics. Rather than forecast each question on an ad hoc basis, I develop guides for particular domains.

My Forecast Factors Series aims to share my own guidelines and forecasts to support more accurate aggregate predictions by the Metaculus community.

The average number of children born per mother varies greatly from country to country, from 6.8 in Niger to 0.8 in South Korea. Birth rates directly alter a country’s population pyramid, and therefore the proportion of working age and dependent citizens. This in turn has major effects on a country’s economy. A population that is too old or too young has too many dependents and cannot take advantage of a demographic dividend and experience accelerated economic growth. 

It is important for policy makers to predict future birth rates and potentially enact policies to change them in extreme cases. In the past, countries have resorted to radical measures to do so, but these often have disastrous long-term consequences. China, for instance, is now encouraging couples to have additional children, due to their TFR being too low now. Therefore identifying the factors that lead to high or low birth rates would enable policy makers to come up with ways to course correct without these unwanted side effects. 

I have reviewed relevant research and have collected the factors that I find most useful for forecasting. After I explain each factor, I will make predictions on Metaculus questions and will further explain my reasoning to demonstrate how to integrate these factors into your forecasting.

Factors

1. Countries that are more urbanized tend to have lower birth rates. Of the top 50 most urbanized countries in the world, only Israel, Jordan, Venezuela and Saudi Arabia have a total fertility rate above replacement. By contrast, among the bottom 50, only Vietnam, India, Barbados, Sri Lanka and Liechtenstein have TFRs below replacement. This correlation can even be seen within the United States.
2. It is notable that Israel, Jordan and Saudi Arabia have higher birth rates than other highly-urbanized countries, since they are all in the Middle East. These countries tend to be more religious than their urbanized European and East Asian counterparts. This idea was supported in the United States by an article published by the National Library of Medicine.
3. There is a very strong correlation between how developed a country is and their birth rate. These factors should include female education, GDP per capita and the availability of contraception. However, there is a notable number of developed countries in Eastern and Southern Europe with lower birth rates than the highest developed countries.
4. The number of children that women express wanting is strongly correlated to how many children they ultimately have. Only in a select few countries does fertility overshoot or undershoot the ideal by one child or more. China had success with its anti-natalist policies in the past, though perhaps it worked too well considering their current low TFR.
5. The demographic transition is speeding up in third world countries. Whereas it took England 95 years to go from a TFR of 6 to 3, it only took China 11 years. This could be related to China’s dramatic economic turnaround.
6. Birth rates tend to decline in times of war. This can be observed during World War 1, World War 2, during the Vietnam War, and Iran and Iraq during their war in the 1980s, among others.
7. Contrary to popular belief, there appears to be little correlation between the strength of a county’s child care system and its birth rate. Countries where women perform a disproportionate amount of the housework tend to have lower birth rates than economic peer countries, suggesting that additional direct assistance to mothers encourages births more than financial incentives. This correlation appears to be minor, however.

Selected Predictions

I found this question to be simple to predict once I had addressed Africa. For Africa, I went against popular projections of fertility declining slowly. Instead I looked for a model that more closely resembled the actual pace of decline in Africa. The closest proxy I found for the continent, which projects a continued pace in TFR decline to 1.8 by 2060, is Jamaica. If Jamaica had not had a large proportion of its population emigrate, it would have hit a peak population roughly double what it was in the mid-1970s, when its TFR was the same as Africa's today. I added another 1.4 billion to the world population, although one might expect the decline to be more heavily focused in East Africa than West Africa.

I found the consensus around India to be more accurate. I believe this is partially due to India being easier to study than countries in the interior of Africa, and partially because it’s relatively normal current TFR. In any case, it should peak at around 200-250m (million) more people than now before settling back down to roughly where it is now. The decline will be sped up by India currently having more kids than desired despite being around replacement. Pakistan should add another 100m at peak due to a much higher current TFR. That’s ~300m more from South Asia.

Europe, the USA, Canada, Australia and New Zealand are all undershooting their desired fertility and have the financial means to increase it, so I would agree that they should reach stable birth rates late in the century. Out of a combined population of ~1.2b, I would guess that they would lose ~2-250m before stabilizing. A good proxy might be ethnic Norwegians in Norway, whose population has remained stable since 1980, when TFR dropped below two.

China is an interesting case. The birth rate suddenly and dramatically crashed in just six years. But there is a decent chance that the birth rate will be either naturally or forcefully increased so as not to fall below 1b people. There is a lot of prestige riding on it and ~13-14m births annually is all it would take to maintain it due to Chinese life expectancy. I'm going to subtract 400m. 

That just leaves Central Asia and the Middle East. Combined they have 550m. With a TFR of ~2.8 and dropping, we can add maybe 200m. Paradoxically, the TFR in Iran fell dramatically after the religious-in-nature Iranian Revolution, and it didn’t recover even after the end of the Iran-Iraq War. Therefore it might be likely that the Middle East might have a TFR around or slightly above Europe’s, as seen in Qatar and nearby countries. Central Asia, which is more secular, is continuing to defy the odds by increasing its TFR after nearing 2, which is unprecedented.

Of course, not all of these peaks and lows will coincide, but all signs seem to point to peak human population occurring around the 2060s, with Africa being the main driver.

Unlike most of their contemporaries, Hungary’s birth rate appears to be trending up and is also one of the developed countries that is most undershooting its desired fertility. I would be surprised if its birth rate wasn’t higher in a decade.

I believe the community is dead-on with their predictions. With the war and economic sanctions that Russia is facing, I find it hard to believe that its TFR won’t drop, despite the Duma’s best efforts.

There is some evidence that Chinese college students desire a very low number of children. If this is indeed true, the Chinese government has its work cut out for them in reversing this trend. The community has little reason to believe that China will begin forced reproduction or will ban abortion and contraception in the next few decades.

Potential Issues

I may have selected factors where correlation is being mistaken for causation. It’s possible that there are hidden variables that more directly affect birth rates than the selected factors, and would ultimately be more useful if identified. It might be that ages at first marriage might be more informative than any economic well-being indicator, for example.

The factors that I selected might be difficult to quantify. For example, one country might self-identify as being more religious than another, while not actually exhibiting any more religious behaviors.

While still being a useful tool, there are flaws with the Total Fertility Rate concept. TFR assumes that women will continue to have children at the same ages as previous generations. This idea has proven false by advancing maternal ages. Therefore TFR might underestimate actual birth rates and the number of children a woman will actually have in her lifetime.

---

Here are some useful links to further research this topic:

• The Demographic Outlook: 2023 to 2053
• Population Division: World Population Prospects 2022
• Projecting age-specific fertility rates by using time-series methods
• The Long-Term Decline in Fertility—and What It Means for State Budgets