Are Counterfactual Population Projections the Gold Standard for Assessing the Demographic Determinants of Population Ageing?

Michael Murphy, LSE

We apply counterfactual population projections to 11 European countries with long-run demographic data series and undertake detailed analysis for three countries, England & Wales, France and Sweden, with base years covering the extended period 1850 to the present to estimate the performance of counterfactual population projections to estimate the contributions of fertility and mortality to population ageing.

We conclude that there are a number of substantive and technical weaknesses in the counterfactual projections approach, in particular the results are very sensitive to the choice of base year. The level of fertility in the base year appears to be the main factor that determines whether the counterfactual projection approach suggests that fertility or mortality id the main driver of population ageing.


Counterfactual projections, which conventionally fix fertility or mortality at an initial value and compare projection results with later actual values, are sometimes used to identify the contributions of fertility and mortality to population ageing. Extended timescales are necessary to mitigate the effect of initial population structure on estimates of population ageing. However, conclusions based on comparisons over long timescales will be with unrealistic projected populations tending to zero or infinity. If the effect of fertility change on population structure is assessed by a projection with fertility fixed at around 1850 levels, the projected populations for the countries considered here would be about eight times as large as the actual population by 2015 and it would have a very young age structure. On the other hand, if mortality is fixed at 1850 values then the projected population would be about one quarter of the actual size. With shorter projection periods, an additional problem is that the results can be very sensitive to the choice of the starting date. For example, if such an exercise was undertaken over a period of about 40 years in Britain starting from the mid-1960s, baseline fertility (TFR in England & Wales of 2.93 in 1964) was much higher than the average over the next four decades. The current observed population structure would be considerably older than the projected population, leading to the interpretation that fertility change in recent decades had made the population older. On the other hand, if the same analysis had been started 30 rather than 40 years ago (TFR 1.66 in 1977, so lower than the average over subsequent period) the actual population would be younger than the projected one, and the interpretation would be that fertility change in that period had made the population younger.

The relative importance of fertility and mortality to population ageing in recent decades in these countries seems to depend crucially whether we undertake the same analysis over the past 95 or past 85 years. The reason for the apparent strong fertility effect over the full twentieth century arises because base year period fertility was considerably higher than average subsequent fertility, and in any case period fertility measures are highly sensitive to timing changes so are a very poor basis for making long-term comparisons.

We have analysed the counterfactual population projections method systematically across a group of high-income countries with different base and final years. While it has been argued that this approach provides the definitive method for establishing the primacy of fertility change as the driver of population ageing in such countries over the twentieth century, and sometimes as a generalisation that holds more generally across time and space, we conclude that this approach has a number of limitations:

  1. The results are highly sensitive to base year distributions and vital rates. A consequence is that apparently minor changes in choice of base year can lead to completely reversed conclusions about the determinants of population ageing.
  2. The base year level of fertility appears to be the main factor that determines whether the counterfactual projection approach suggests that fertility or mortality is the main driver of population ageing.
  3. The results are not transitive across time, i.e. the effects of fertility and mortality measured in sub-intervals do not add up to impact over the whole interval.
  4. These is no clear index for attributing changes to fertility or mortality (one possible index is derived and presented).
  5. Different ageing indices tend to produce similar qualitative conclusions, but quantitative results may differ markedly.
  6. The usual approach does not distinguish the effects of initial age structure and subsequent fertility and mortality rates.

Use of a small analysis window at the end of an expanded influence period mitigates some, but not all, of these problems.

Counterfactual projections and stable population modelling have been used to argue for the primacy of fertility change on population ageing across high income countries over the past century or so. However, some of the early work in this area was much more nuanced. Kisker (1950, p. 57) concluded: “As for the relative importance of these factors [fertility, mortality and migration], much depends on the population and period of time under consideration”, and Thompson (1948) in the middle of the twentieth century stated although declines in US fertility were historically the most important, especially in the period 1920-50, in future declines in mortality may become the most important. This would appear to have been an accurate prediction and current trends suggest that mortality will continue to remain the primary driver of population in high income countries in the twenty-first century, with historical legacy the most important in the rest of the world, apart from Least Developed Regions where future fertility trends will remain crucial.

Presented in Session 1065: Data and Methods