The size of the human population is one of the most important metrics to understand the world. This is true globally, regionally, nationally, and at local levels. Total numbers are often meaningless without the context of population figures.

This doesn’t just apply to population sizes today: we also want to understand what populations might look like in the future. We need it to allocate resources and budgets, and to understand health, socioeconomic, environmental, and other demographic trends.

That means accurate population projections are very valuable. They let us plan for the future.

The United Nations (UN) publishes projections in its World Population Prospects. It has been doing this periodically for more than 50 years.

But how much should we trust these projections?

One way to gauge the credibility of UN projections for the future is to look back at its track record of predictions in the past.

In this article, I compare the UN’s historical projections to the latest population estimates to see how close they were.

Most of the UN’s historical population projections have been close to final estimates

Many of the UN’s historical projections – even as far back as the 1960s – have been remarkably close to the truth.

Every few years, the UN publishes a new report on population trends, which it calls a “revision”. The report published in 2022 is called the “2022 revision”; the one from 1990 is the “1990 revision”, etc.

In the chart, I’ve shown a range of these historical revisions, with their projections for the world population between 1970 and 2020. The UN’s latest population estimates – which is our closest estimate of the true population – are also shown for comparison.

If you hover over the ‘latest observed’ data label on the chart you can see how it compares to the various projections around it.

As you can see, most historical projections were close to the final figure, even when projecting three to four decades in the future. We’ll take a close look at the exact figures below.

Most projections differed from the latest estimates by just 1% to 2%. Many were within 1%, and none differed by more than 5%.¹ This is not bad, especially considering that global estimates of the world population today can vary by around 1% between sources.²

We can also take a closer look at how these historical projections stack up. In the table, I’ve provided the latest population estimates from the UN in the top row (so you can see the global population in 1980, 1990, 2000, etc.).

Below are the exact numbers that each UN population revision projected for these same years. So you can see that the 1968 revision projected that the world population in 1980 would be 4.46 billion. The actual number was 4.44 billion. So, pretty close. But further out the difference becomes considerable — for the year 2000 the demographers back in the late 60s overestimated the world’s population size.

You can see that the projections across the years did vary, but most clustered within a small range and came pretty close to the final estimate.

Global projections have been relatively accurate, but this isn’t true for every region

We should not assume that because global projections have been reasonably accurate, projections for individual countries or regions have been equally so. This also applies to specific age groups.

Over- and under-estimates of particular populations canceled each other out, leading to a total figure that seemed more accurate than it was for individual groups. For example, an overestimate of the population in Asia could be partly canceled out by an underestimate in Europe.

In a paper published in Population Studies, the researcher Nico Keilman, studied the accuracy of UN population projections by looking at the errors of large regions, age groups, and specific estimates of fertility and mortality trends.³

To give one example: in Europe, UN projections would often overestimate the number of future children in a population but underestimate the number of elderly people. This was because the UN expected birth rates to be higher than they turned out, and it was at the same time too pessimistic about improvements in life expectancy (especially for women).

Another significant error for some regions – affecting mostly Africa, Asia, and Latin America – was inaccurate population estimates in the first place (in what we would call the “base year” when the time-series starts). Many countries in these regions had poor census records and national statistics on the size and structure of the population. If the population in the start year is wrong, the projected population could be very wrong, even if you make reasonable assumptions about changes in fertility and mortality trends.

Keilman notes that the quality of population data has improved over time, so this inaccuracy got smaller over time.

Long-term projections are less accurate

Global population projections often perform well when we compare them to reality 5, 10, or 20 years into the future. But over longer timescales, they often diverged from the real pathway.

See the chart below, which extends the historical projections further into the future – to 2030. You can see that in earlier years, there was much less variation between them. But with time, they start to diverge.

This makes sense: to make projections about the population, researchers need to make assumptions about how fertility rates – the number of children the average woman has – and life expectancy will change. And researchers need to do this across the world, with very different starting points and socioeconomic futures. Small errors in these assumptions at the start of a trend will accumulate and turn into much larger ones.

We can understand the challenges in making long-term projections by looking at the most recent UN projections. In its 2019 revision, the UN projected that the world population would not peak by the end of the century: it expected numbers to reach 10.9 billion by 2100 and continue to grow slowly in the early 22nd century. Its 2022 revision however made quite a different projection: now the UN expects that the world population will peak at around 10.4 billion people in the mid-2080s.

This update, of course, came from an improved understanding of how fertility rates and other demographic indicators were changing across the world. But it highlights how hard it is to predict these changes for every country, 50 to 100 years into the future.

To summarize, short-to-medium-term projections have often been close to the true figure. But the global aggregates can hide errors on the national and regional level and we should be cautious about relying too heavily on long-term ones.

Hannah Ritchie (2023) - "The UN has made population projections for more than 50 years – how accurate have they been?". Published online at OurWorldInData.org. Retrieved from: 'https://staging-owid.netlify.app/population-projections' [Online Resource]

@article{owid-population-projections,
    author = {Hannah Ritchie},
    title = {The UN has made population projections for more than 50 years – how accurate have they been?},
    journal = {Our World in Data},
    year = {2023},
    note = {https://staging-owid.netlify.app/population-projections}
}

Reuse this work freely

All visualizations, data, and code produced by Our World in Data are completely open access under the Creative Commons BY license. You have the permission to use, distribute, and reproduce these in any medium, provided the source and authors are credited.

The data produced by third parties and made available by Our World in Data is subject to the license terms from the original third-party authors. We will always indicate the original source of the data in our documentation, so you should always check the license of any such third-party data before use and redistribution.

All of our charts can be embedded in any site.