Population forecasts have a poor track record. For much of the 20th century demographers were predicting doom: either population would explode, or it would collapse; either there would be too many young people, or not enough; too much work to do, or not enough to go round.

None of these scenarios materialised. The population forecasts usually proved wrong; and even when they were right, market economies were flexible enough to adjust to demographic circumstances.

Currently the great scare is about the ageing population. The poor struggling taxpayers will be crushed under the burden of old people, as life expectancy increases, and not enough children are being born to maintain the workforce. In more technical language, the concern is that dependency ratios will rise to unacceptable levels.

I commented on all this a year or two ago, and on re-reading my old post I find little to alter. But some recent discussion on another website has prompted me to do a bit more delving into the statistics. [See note 1 for some useful online sources.]

First, the meaning of dependency ratios. As usually defined, the dependency ratio (DR) is the ratio between the non-working-age population and the working-age population itself (conventionally taken as aged 15-64 inclusive), usually expressed as a percentage. For example, if 10% of the total population is aged 0-14, and 20% is aged 65 and above, then the working age population is 70%, and the DR is [(10+20)/70] x 100 = approx. 43%. In principle, the DR can exceed 100%, though this is rare.

Here are some DR figures from various sources. The figures after 2005 are projections, and therefore not to be entirely trusted, though figures for 2025 are unlikely to be badly wrong:

...............2005 2025 2050
USA......... 51.... 62
UK............52.....59......70
Japan...... 47..... 67..... 87
Germany.. 49...60.....77
France..... 54....64..... 76
World ave. 58...53
SSAfrica.....89...72
LatAmer....59...49


A few points stand out from this table:

a. The major developed countries at present have similar DR’s, slightly above or
below 50%.

b. Contrary to common assumptions, DR’s in Japan and Germany are currently lower than in other developed countries.

c. DR’s in less developed countries are relatively high, especially in Sub-Saharan Africa.

d. DR’s in developed countries are expected to rise over the next 50 years, while DR’s in less developed countries are expected to fall.

The DR reflects previous population history as well as current birth and death rates. A ‘bulge’ of births like the post-war Baby Boom has very long-term effects. Initially the DR is high because there are a lot of dependent children. When the birth rate falls, the DR also falls. But if the low birth rate continues, and life expectancy is high, the DR will eventually rise again as the ‘bulge’ cohorts reach retirement, before falling again as they die off. There are also ‘echo’ effects when the bulge generation itself has children, and so on. In developed countries the post-war Baby Boom generation is now reaching retirement age, so DR’s are rising. The increase in the DR is expected to be especially sharp in Japan, as a high birth rate in the 1960s was followed by a sharp downturn in the 70s. As I pointed out in my previous post, Japan (and to a lesser extent Germany) are facing a triple whammy: low birth rates, high life expectancy, and a large ‘bulge’ working its way through.

But how serious is all this? What level of DR is ‘acceptable’?

Consider a very simplified population model. Suppose each woman has 2 children; infant mortality is zero; and everyone lives to the same fixed age and then dies. Suppose also that the same pattern has always existed and there is no immigration or emigration. The population will therefore be distributed evenly over all age groups up to the fixed age of death, and the DR will be determined by this age. For example, if the fixed death age is 65, the DR will be 15/50 = 30%, if it is 70 the DR will be (15+5)/50 = 40%, if it is 80 the DR will be (15+15)/50 = 60%, and if it is 90 the DR will be (15+25)/50 = 80%.

Although this is a very simplified model, it gives an approximation to what we might expect in developed countries. We expect a reasonably stable population, we want low infant mortality, and we want to live to a ripe old age. In the developed world infant mortality is already very low, and life expectancy at birth is between 70 and 80. Most deaths occur in old age. From the simple model we might therefore expect current DR’s to be between 40% and 60%. If average life expectancy rises to 90, which is probably around its natural limit, the DR would be approaching 80%. [Note 2]

DR’s in less developed countries are generally higher. In traditional agricultural societies infant mortality is high, but women have many children. There are relatively few old people, but at any one time there are likely to be about as many children as adults alive, so DR’s are about 100%. This pattern is still often seen in sub-Saharan Africa.

An intermediate pattern can be seen in early-modern England. Between 1540 and 1850 the DR never fell below 60%, and in the early 19th century rose to about 80%. [Note 2] By 1901 it had fallen to 60%. It declined to a low point of 43% in the 1930s, but rose again to 59% in the 1950s, with the Baby Boom, and has been falling since then, to its present level of 52%. [Note 3]. I don’t have long-term figures for other countries, but in all the major developed countries the DR has fallen since the 1960s. Dependency Ratios in developed countries are now at historically low levels.

The prevailing DR of around 50% is the result of unique historical circumstances, and we should not necessarily expect it to be sustained. In particular, attempting to maintain it by encouraging high immigration would be short-sighted. As immigrants are usually young adults, the immediate effect would be to reduce the DR, but it would soon increase again as the immigrants had children (especially as they tend to have high birth rates). Eventually the immigrants would themselves grow old, and DR’s would be back roughly where they started, but at a higher level of population. In densely populated countries like Japan, the UK, and the Netherlands this would cause serious problems. Even at present rates of immigration and population growth in the UK, the countryside is rapidly disappearing under concrete.

Of course, the DR is based simply on the numbers of people in given age groups, and does not tell us how many people are actually working. The following table, based on ILO figures, shows selected economic activity rates. The EAR shows the proportion of people who are ‘economically active’, which means working or seeking work in the labour market. It therefore excludes those who are in full time education, retired, disabled, housewives not in the labour market, etc. The ILO figures show the proportion of economically active people among all people aged 15 and above, including the elderly. (EARs relative to the entire population, including under-15s, would be lower.)

...................1950 1970 2000
USA............ 43..... 43.... 51
UK.............. 46..... 46.... 50
Japan......... 44...... 51....54
Germany... 48...... 46....50
France........ 46..... 43... 45

It will be seen that these EARs are all within a fairly narrow range. Except in France they have slightly increased between 1950 and 2000. Since the proportion of children (not counted in these figures) has also fallen, the number of ‘dependants’ supported by each ‘worker’ has in fact fallen substantially. Yet over this period the average length of full time education has increased, average retirement ages have fallen, and life expectancy has increased, all of which might be expected to reduce the EAR. So what is going on? Part of the answer may be that the post-war baby boom is distorting the EAR as well as the DR. But the other obvious factor is that the proportion of women who are economically active has increased, for example from 24% to 48% in the case of the US. A lower birth rate means that more women can work.

I have not seen any projections for EARs beyond 2010. What is clear is that there is still scope for EARs to increase. Elderly (but healthy) people are capable of doing some paid work. Even a reversal of the trend towards early (pre-65) retirement would make a significant difference to EARs. For example, in Germany at present less than 30% of men and 10% of women between 60 and 65 are still working. I believe it will be possible without major disruption to maintain EARs at around 50% throughout the period to 2050, [note 5] after which the problem of the elderly ‘bulge’ will begin to ease. So there is no need for panic.

Of course, part of the concern about the ‘ageing population’ is not just about the number of old people, but the cost of supporting them. It seems to be assumed that old people, on average, are a great burden on the taxpayer, and far more expensive than children.

The question of economic transfers between generations raises conceptual, ethical, and statistical problems. I may devote another post to these, but here I will say only that the cost of supporting old people is often overstated. It is not legitimate to assume that the cost of public pensions falls on current taxpayers. This is only justified to the extent that pensions have not been adequately funded by the contributions of the pensioners during their working life, whether explicitly though pension schemes or through equivalent general taxation. So far as the costs of health and social services to old people are concerned, I have been looking at the figures for the UK, and I find that the average cost per head of old people is considerably less than the cost of public expenditure on children and young people. Fifteen years or so of education does not come cheap. But this is a big subject and I may return to it.


Note 1: For European figures see here source

For Japan see here: source.
For the USA and global comparisons see here:
source
For ILO data on EARs see here: source

Note 2: As mortality in old age is bound to be skewed, the median age at death will be somewhat below the mean, so the DR will not quite reach the level implied by mean life expectancy.

Note 3: Based on E. Wrigley and R, Schofield: The Population History of England, 1541-1871, p.443-5. Wrigley and Schofield use a threshold of 60 rather than 65 for old age, so I have deducted 5% from their estimated DR’s for comparability with other figures.

Note 4: Based on D. Coleman and J. Salt, The British Population, p.545.

Note 5: for example, in Japan the projected 2050 DR of 87% corresponds to a population of about 57m aged 15-64, 36m aged 65+, and 14m aged 0-14. With these proportions it would require about 80% of the working age population to be economically active to produce an EAR of 50% (on the ILO basis). This could be achieved for example by an activity rate among the working-age population of 90% for males and 70% for females, which does not seem unfeasibly high. It would allow both men and women to spend on average 5 years in post-15 education, and women to spend a further 10 years in childcare.