Population projections: why they are often wrongBMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g5184 (Published 20 August 2014) Cite this as: BMJ 2014;349:g5184
As an economist it is always nice to come across another profession that, for many good and legitimate reasons, struggles to always make accurate predictions of the future; hello demographers. Predicting the size and composition of future populations is a fundamentally important thing to do. Population projections underpin our estimates of future healthcare needs, government spending and tax revenues (and possible debt and deficits), housing demand, and road, rail, and air transport needs. The list isn’t endless, but it’s certainly long. It’s perhaps unfortunate then that it turns out that population projection is quite a difficult thing to do and one which we repeatedly get wrong.
Demographers who construct population projections know how difficult it is to get these right. The United Nations makes population projections for the world and its countries up to 2100 based primarily on assumptions about future mortality, fertility, and international migration rates.1 These suggest that, for example, the populations of the more developed countries will shrink as a proportion of the world’s population from around a third to just over a tenth by 2100. But, as fig 1⇓ shows, they also show the enormous range of the overall projections depending on assumptions made about the future. The world’s population could increase by 125% over the next 75 years, or perhaps peak at around 2050 and then shrink by 7% or so over the next half century, or maybe not.
For the UK, the Office for Budget Responsibility (OBR) has usefully pointed out just how wrong population projections can be.2 Revisiting projections made from the mid-1950s through to the latest in 2012, the OBR shows that, in general, compared with actual counts of the population (and estimates based on censuses), past projections tend to have underestimated total population numbers and (as might be expected) to get them more wrong the further forward the projections go (figs 2⇓ and 3⇓). The 1965 based projections were an aberration and the most wrong of all the projections so far—assuming that the high birth rates of the 1960s would continue. The most successful seem to be those made in 1985, being pretty much spot on up to 2001 and then veering off to an underestimate of 4.2 million (around 7% of the actual population) by 2012.
So why are the projections generally wrong? Leaving aside problems with actually counting people to get an accurate baseline to make projections, populations change for three reasons: births, deaths, and migration. Predicting how these will change has proved hard. OBR’s analysis suggests that net migration (fig 4⇓) has been particularly hard to predict, and for projections made from 1977 to 2004 migration estimates accounted for the majority of the error in the projected population for 2011. Also, earlier projections of fertility rates tended to overestimate births whereas later projections underestimated rates.
What is particularly striking is how consistently wrong projections of deaths have been— and all in the same direction, overestimating the number of deaths (fig 5⇓). The 1975 based projection, for example, overestimated the number of deaths in the UK in 2011 by 132 000— nearly a quarter of the actual number. All mortality projections show a turning point where the number of deaths starts to rise, reflecting the 1950s-60s baby boomer bulge working its way through the population. But that turning point has moved with each projection.
Future projections will inevitably remain uncertain. As the OBR does, the best thing is to recognise this and construct various alternative futures.
Cite this as: BMJ 2014;349:g5184
Competing interests: I have read and understood BMJ policy on declaration of interests and have no relevant interests to declare
Provenance and peer review: Commissioned; externally peer reviewed.