Page 101 - IJPS-10-4

International Journal of
Population Studies Mortality shapes population age structures

time. For this reason, our model is not very effective in age population is large and the dependent population …
predicting short-term changes in age structures, and is small.” Our proposition is that a comparison should be
analyses of the determinants of such changes (and notably made with the standard (stationary) age structure. In this
those on the causes of population aging) may result in sense, a demographic dividend (or the “first demographic
the opposite conclusion that fertility matters more than dividend,” in Lee and Mason’s words) can be considered
mortality. In the simplest terms, these analyses involve present as long as the observed age structure is “better”
regressions of the following type (Equation VII): than that of the corresponding stationary population.
In Figure 2, for instance, the old age dependency index
*
c f c , F, M (VII) of China (OADI, right panels) started to worsen (i.e., to
,
increase) since the 1980s and is currently undergoing rapid
where the deltas indicate variations, and ε is the error deterioration. However, until approximately 2036, the
term. In Equation VII, a variation in the age structure observed OADI in China will be lower than its standard
∆c is thought to be caused by variations in survival, equivalent OADI*. In our interpretation, this means that
fertility, and migration (∆c*, ∆F, and ∆M, respectively), until then, China will still benefit from a demographic
and a decomposition analysis of some sort will attempt to bonus, although this will become a bit smaller every year.
assess the relative role of each of these forces. Short-term The question may be asked in more general terms:
variations in survival, however, are generally modest and When looking at the age structure of a population, what
may even be non-existent; in Equation VII, this may well should observers expect to find and against what implicit
result in a lesser role estimated for survival in determining standard do they asses the current situation? Our answer
population aging. However, if our hypothesis is correct, is that (excluding exceptional cases, such as an ongoing
this is nonetheless consistent with the predominant role war), the best possible expectation should be based on the
of survival in shaping population age structures in the stationary age structure of the period. All relevant aspects
long run. of the observed age structure (peaks and troughs, larger or
This should help explain why several demographers, smaller bases, etc.) can be better appreciated compared to
following Coale (1956; 1957), tend to think that it is mainly this standard.
fertility that causes population aging (Bengtsson & Scott, Our findings clarify some of the issues behind the
2005; 2010; Keyfits, 1975; Lee & Zhou, 2017), although frequently re-surfacing question, “Are immigrants
exceptions for developed countries have been frequently substitutes for births?” (United Nations, 2000). Those who
noted (Fernandes et al., 2023; Murphy, 2017; Preston et al., reject this notion typically take some observed structural
1989; Preston & Stokes, 2012). As mentioned, our claim index (e.g., the OADI or its reciprocal, the support ratio)
is that mortality is the main driver of the shape of age and simulate the number of immigrants needed to preserve
pyramids in the long term, although its action is slow and that value forever—frequently, but not surprisingly, arriving
may remain in the background, especially in times of rapid at implausibly high values (e.g., Coleman [2008]). In light of
change, such as demographic transition. our findings, this question appears to be misleading. The data
Our findings may influence the debate regarding the points outside the bisector (e.g., Figure 5), suggest that the
advantages and disadvantages of cross-sectional life tables. demographic structure is biased (i.e., far from its standard)
Some demographers (e.g., Borgan & Keilman (2019) and and that this situation will not last: things will eventually
Keilman (2019) are skeptical about their theoretical and get back to “normal.” Only standard-consistent structural
practical utility and argue that they are likely to convey targets can be pursued in the long term, and, with this
biased results in periods of mortality change (i.e., almost more modest target in mind, simulations (not shown here)
always). Our study indicates that they are useful both in indicate that immigrants can be quantitative substitutes for
theory and practice. births unless fertility drops to extremely low levels.
In the same spirit, our findings may prompt Finally, it is noteworthy that due to mortality changes,
demographers to reconsider the notion and measurement all target values move over time (notably the OADI of
of “demographic bonus” or “dividend.” Until now, to the Figure 2 [right panel]), unless improvements in survival
best of our knowledge, the “window of opportunity” has are neutralized by offsetting adjustments in threshold ages,
been defined in terms of improvement over the recent such as higher retirement ages. In other words, policy
past. For example, Lee & Mason (2010) call it a “phase measures of this type, while adopted under the pressure
of … three to five decades … [when] support ratios rise of circumstances (budget imbalances in pension systems)
well above their pre-transition levels”, while Crombach & and wearily defended by reluctant policymakers, may find
Smits (2022) describe it as “a period in which the working- theoretical support in our approach, which is consistent

Volume 10 Issue 4 (2024) 95 https://doi.org/10.36922/ijps.377