How accurate are population forecasts?

Answers to your questions about the news.
Jan. 27 2011 6:32 PM

Crowded, With a Chance of Overcrowded

How do demographers forecast population growth?

Crowd. Click image to expand.
Accurate forecasting of population changes depends on good data

The global Muslim population will grow twice as fast as the non-Muslim population over the next 20 years, according to a report from the Pew Research Center and the John Templeton Foundation published on Thursday. During the same time period, the American Muslim population will more than double. How do demographers come up with these numbers?

Using the "cohort-component" method. In almost every serious population forecast, demographers first slice a given population into tranches, such as women ages 20 to 24 and men ages 65 to 69. Then they issue separate forecasts for each, typically using just three variables: fertility rate (how many children women have), mortality rate (what proportion of the population dies per year), and migration. Usually, demographers will calculate a range of scenarios for each variable—for instance, the highest and lowest plausible fertility rates over the next, say, 20 years. Finally, they synthesize the sub-forecasts into a range of likely trends, which they often present as "high," "medium," and "low" estimates.

Advertisement

Population forecasts are generally quite accurate. Country-level forecasts, such as those made by the United Nations and the World Bank, miss the mark by about 6 percent on average. (The U.N. has been projecting population growth at a global and regional level since the 1950s, and for most individual countries since shortly thereafter.) There's been remarkably little change in most demographers' techniques over the past century, so accuracy is mostly a function of how good (and how current) the underlying fertility, mortality, and migration data are. A study from 1983 found that while neither U.N. predictions for developed countries nor the Census Bureau's predictions for the U.S. population had improved, those in developing countries had become more accurate—likely because of more-detailed recordkeeping. A 2001 study found little change in the U.N.'s overall forecasting accuracy since 1950s; if anything, accuracy peaked in the 1970s and late 1980s. (In general, the most difficult continent to make predictions for has been Africa, while Europe has been the easiest.)

Population forecasting has been caught off-guard by a few big trends. The baby boom made demographers' population projections look foolishly low in the wake of World War II. Then, they predicted the boom would last longer than it did; the so-called "baby bust" made forecasts from the mid-1950s look much too high. Demographers also didn't anticipate how severely the spread of HIV through the developing world would increase mortality rates.

It's hard to say whether the Pew and Templeton forecast of the Muslim population will bear out, but the forecast does have a few things going for it, including the fact that fertility, mortality, and migration rates in the U.S. have been well-studied. The Muslim fertility rate is fairly low and has not been very volatile, another factor that should improve accuracy. On the other hand, as the population in question gets smaller—from global to regional to national to religious sub-samples—accuracy tends to decrease. And while the U.S. has a long history of population-counting, our census doesn't ask about religion. (South Africa's census, on the other hand distinguishes among more than 60 faiths.) Instead, the American projections are based on private surveys that institutions like Pew regularly undertake. 

Got a question about today's news? Ask the Explainer.

Explainer thanks Brian Grim of the Pew Research Center's Forum on Religion & Public Life, Carl Haub of the Population Reference Bureau, and Ron Lee of the University of California at Berkeley.

Like  Slate and the Explainer on Facebook. Follow us on Twitter.