1.1. Twice as many over-75s by 2070

Between 2021 and 2070, under this baseline scenario, population increase is concentrated exclusively among the over-75s (Figure 1). It doubles at these ages thanks to rising life expectancy and the replacement of the pre-1946 birth cohorts by the slightly larger cohorts born from the end of the baby boom to the mid-1990s. The population at ages 60–75 remains relatively stable, with a slight increase for men and a small decrease for women. The sex ratio at these ages becomes less skewed in favour of women, although women still represent more than half of the total. Below age 60, on the other hand, the population falls by 10%.

Accompanied by 26 alternative scenarios combining different fertility, mortality, and migration assumptions, the baseline scenario is based on a total fertility rate of 1.8 children per woman (a level similar to that of today), a mortality decline that continues at the same pace as in the 2010s, and net migration (arrivals in France minus departures) of +70,000 persons per year. Under the ‘high’ and ‘low’ fertility scenarios (2.0 or 1.6 children per woman), the population in 2070 varies by plus or minus 4 million around the 68 million of the baseline scenario. Likewise, under the high and low net migration scenarios (+120,000 or +20,000 per year) it varies by plus or minus 4 million. The alternative mortality scenarios have a more limited impact on population size: plus or minus 2.4 million. Combining the high or low assumptions of each component, the population projections range between a minimum of 58 million and a maximum of 79 million in 2070 (Figure 2).

The degree of uncertainty varies by age (Figure 3). It is highest for the cohorts not yet born in 2021. For example, the number of persons aged 0–19 in 2070 is 14 million under the baseline scenario, but the range of results is broad: between 11 million for the low scenario and 17 million for the high scenario, i.e. 25% above or below the baseline projection. For persons aged over 75 on the other hand, already born in 2021, the extreme scenarios produce figures of 10 million to 14 million in 2070, i.e. 18% above and below the baseline projection of 12 million. The projected number of people aged over 75 between 2021 and 2070 is affected mainly by the mortality assumption. Whatever this assumption, their number increases very sharply (by 89% in the baseline scenario), as life expectancy increases and the cohorts born between 1969 and 1994 replace the depleted cohorts born before 1946.

1.2. Greater uncertainty for births than for deaths

The baseline scenario projects that up to 2070, annual births will decrease by 80,000 (–10%) compared with current levels (740,000 births in 2020; 660,000 in 2070) (Figure 4a). Annual numbers of deaths will increase by 160,000, however, rising from 610,000 in 2019 before the COVID-19 pandemic to 770,000 in 2070, an increase of 25% (Figure 4b). These figures differ from one scenario to another, with a larger variation for births than for deaths. In 2030, the number of births ranges from 610,000 to 810,000 between the low and high fertility scenarios (Figure 4a). In 2070, the range is between –190,000 and +230,000 annual births with respect to the baseline scenario (660,000). The annual number of deaths increases rapidly in all cases (Figure 4b). For 2070, the range of plus or minus 50,000 deaths with respect to the 770,000 of the baseline scenario is much smaller than that of births. All scenarios lead to a substantial increase in deaths between now and 2045, as the large baby boom cohorts reach very advanced ages and gradually die out.

Under the baseline scenario, from 2035 onwards, deaths outnumber births and natural growth becomes negative, reaching –110,000 in 2070 (Figure 5). This figure varies widely across scenarios, from –330,000 to +160,000, but net migration remains positive in all cases (between +20,000 and +120,000 per year), so in 11 out of 27 scenarios the population is still growing in 2070.

1.3. Looking back on 50 years of changing projections

How have the assumptions used in the INSEE projections evolved over 50 years? Here, we will examine those of the baseline scenario. Life expectancy assumptions have changed considerably over the last half century (Figure 6) [4, 5].

Life expectancy at birth for both sexes combined was 72 years in 1970 (compared with 82 years in 2021), and it was believed that little further progress could be expected. After increasing sharply in the years following the Second World War and gaining almost 5 months annually in the 1950s, life expectancy rose more slowly in the 1960s, giving credence to the notion of a biological limit (Figure 6).

In the 1970 projection up to 1995, life expectancy continues to increase, initially at the same pace as in the 1960s and then more slowly. The following projection, published in 1979, is more pessimistic, with an even slower rise in life expectancy that levels off at 73.8 years in 2000 for both sexes combined. These two projections severely underestimated the true rate of progress. The 1986 projection corrected for these discrepancies by anticipating a much faster rise in life expectancy while continuing to assume it would slow down and level off over the long term.

Despite this adjustment, life expectancy again progressed more quickly than projected, and the expected ceiling was reached and surpassed in 1997. Observing that even the raised life expectancy ceilings were regularly exceeded after a few years, subsequent projections abandoned the idea of a biological limit. Recent trends are extrapolated across the entire projection period, with no upper limit, and this corresponds quite well to the trends observed up to now.

Fertility projections, by contrast, have changed little over 50 years (Table 1). Apart from the projection published in 1970, which forecast much higher fertility than the others (the baby boom was still ongoing at the time, though reaching its end), revisions have been minor, each time reflecting recent upward or downward changes in the fertility rate. While stable since the end of the baby boom, this rate has fluctuated between 1.7 and 2.0 children per woman on average.

Assumptions about net migration have changed significantly over the last 50 years. Projected levels remained high in 1970 (130,000 per year) but dropped sharply in the two following projections (published in 1979 and 1986) after the government decision to end labour migration in 1973. Net migration then varied by a factor of 2 in subsequent projections, between 50,000 and 100,000 per year. A 50,000 difference in net migration affects long-term population growth to the same extent as a difference in the total fertility rate of 0.2 children per woman (see Figure 2).

***

What do these new projections tell us? They confirm that population ageing will continue until 2040, followed by a likely slowing of its increase [1]. Although the projected size of the population in 2070 strongly depends on the choice of assumptions, all scenarios predict a large rise in the proportion of old people up to this time horizon. The extent of variations between scenarios and the effects of adjustments to the baseline scenario point up the uncertainties associated with long-term population projections, while confirming the inevitable process of population ageing. The new baseline projection indicates that a slow decrease in population from 2044 onwards is the most likely future scenario if current trends persist.