I drafted the attached paper for the Journal of Economic Perspectives (the AEA’s “general” journal that has non-technical articles any economist can read) as part of a planned (virtual) symposium on fertility.
The very low levels of fertility in nearly all of the rich, industrial world is drawing increasing media and popular attention, for reasons on two very different time scales.
The very long-run. the fall in fertility to roughly (or below) the replacement rate of 2.1, first in rich countries and now pretty much globally except for Africa and a few other places, implies the “population bomb” scare of the late 1960s and 1970s has now reversed. All of the projections are now that global human populations will peak in a foreseeable future and then start a decline. And, as a new book by Dean Spears and Michael Geruso After the Spike: Population, Progress, and the Case for People points out, if fertility does not get back above replacement levels, the only stopping point for human population is zero. Nick Eberstadt has talked about the “de-population bomb.”
The medium-run (decades). But the challenge facing most countries over the next few decades is not de-population or even so much that their population will fall, but that the age composition of their population will shift towards older people. Over the coming decades in countries and regions that had an early fall in fertility (e.g. before 1990) to near or below replacement the absolute number of old is going to grow and the absolute number of young is going to shrink. So the ratio of young to old is going to fall to levels never before observed in human populations. My new (draft) JEP paper is about that.
As Larry Summers has said “The special sauce of economics is arithmetic” which means getting not just the direction of change rate but magnitudes. So this paper is mostly demographic arithmetic of facts and forecasts. There are four key points.
First, fertility has fallen to near or below replacement in every region but Africa and (and this surprised me) a very simple regression of country level TFR on three country factors explains all of that fall. That is, if one does a simple OLS descriptive regression of TFR across countries and time (from 1950 to 2020) on just (a) years of schooling of the female 15-64 population, (b) child (under 5) mortality rate, and (c) GDP per capita (in PPP) then this explains 85 percent of the total variation. Moreover, once one accounts for those factors there is no unexplained secular trend. Of course these three factors have both complex chains of causal effects and are associated with lots of other changes in economy and society, but the changes in fertility not captured by levels and trends in these three factors are pretty small. I was quite surprised by that and don’t want to make too much of it, but don’t want to make too little of it either.
Second, Africa (both Sub-Saharan and North) and South Asia account for more than 100 percent of the forecast increase in labor force aged (15-64) population to 2050. It is more than 100 percent as the “early fertility transition” regions are going to have fewer youth. The rough numbers are:
The three early fertility transition regions (Advanced, Democratic, Rich Industrial (ADRI), China, FSU/EE) are going to have 355 million less labor force aged people.
Late, but near replacement TFR now, regions (West Asia, Southeast Asia and Latin America and Caribbean) are going to add 190 million.
So the net change in LF aged population to 2050 of those countries is 165 million less.
South Asia, though it is also near replacement in 2025 is going to add 370 million LF aged.
Africa is going to add 800 million.
So the net increase in labor force is going to be about 1 billion (800 plus 370 plus 190 less 355) and Africa is going to be 80 percent of that.
A basic demographic and economic fact about the next few decades is that in the highest productivity countries there are going to be substantially fewer people labor force aged and in the lowest productivity countries there are going to be many, many more labor force aged people.
Third, to get from demographic scenarios (labor force aged) to economics (labor force) and from there to the ration of labor force to population over 65 (with people over 65 potentially in both categories if they are working) we need, naturally, a few more assumptions. I use the ILO data on labor force participation by five year age categories and sex to calculate the 2020 labor force to population 65+. If we assume that, for each country, the LFPR by age and sex remain constant then we can calculate the LF/65+ (again, this does not assume all those 65+ are out of the labor force) for each ADRI country.
These numbers reinforce that many countries are headed into, and far into, economic conditions never before seen in human history. In Italy and Spain the zero-migration, constant LFPR rate, 2050 scenario is less than one worker per person over 65. In Japan and South Korea the ratio will be 1.2 workers. Given that at 1.8 workers per person 65+ the Japanese government is already facing an ageing crisis and ramping up the recruitment of foreign workers to meet labor needs, it is not clear and economic future without much, much, larger scale availability of foreign workers is feasible.
Fourth, the scale of the demographic and labor gaps are too large to be met by existing modalities of movement and hence much, much, larger use of modalities of labor mobility that are contractually limited (e.g. time limited, sector limited) will be needed to balance the economics of labor scarcity and the politics.