DEMOGRAPHIC RESEARCH
VOLUME 35, ARTICLE 29, PAGES 867
−
890
PUBLISHED 22 SEPTEMBER 2016
http://www.demographic-research.org/Volumes/Vol35/29/
DOI: 10.4054/DemRes.2016.35.29
Research Material
Visualising the demographic factors which shape
population age structure
Tom Wilson
©2016 Tom Wilson.
This open-access work is published under the terms of the Creative Commons
Attribution NonCommercial License 2.0 Germany, which permits use,
reproduction & distribution in any medium for non-commercial purposes,
provided the original author(s) and source are given credit.
See http:// creativecommons.org/licenses/by-nc/2.0/de/
Contents
1
Introduction
868
2
The components-of-change population pyramid
870
3
More examples
874
3.1
England and Wales
875
3.2
Scotland
875
3.3
Finland
876
3.4
Sweden
876
3.5
Switzerland
877
3.6
Queensland
877
3.7
Northern Territory
877
3.8
Quantifying the effect of net migration on age structure
878
4
Conclusions
886
References
888
Demographic Research: Volume 35, Article 29
Research Material
http://www.demographic-research.org 867
Visualising the demographic factors which shape population age
structure
Tom Wilson
1
Abstract
BACKGROUND
The population pyramid is one of the most popular tools for visualising population age
structure. However, it is difficult to discern from the diagram the relative effects of
different demographic components on the size of age-specific populations, making it
hard to understand exactly how a population’s age structure is formed.
OBJECTIVE
The aim of this paper is to introduce a type of population pyramid which shows how
births, deaths, and migration have shaped a population’s age structure.
DATA AND METHODS
Births, deaths, and population data were obtained from the Human Mortality Database
and the Australian Bureau of Statistics. A variation on the conventional population
pyramid, termed here a components-of-change pyramid, was created. Based on cohort
population accounts, it illustrates how births, deaths, and net migration have created the
population of each age group. A simple measure which summarises the impact of net
migration on age structure is also suggested.
RESULTS
Example components-of-change pyramids for several countries and subnational regions
are presented, which illustrate how births, deaths, and net migration have fashioned
current population age structures. The influence of migration is shown to vary greatly
between populations.
CONCLUSIONS
The new type of pyramid aids interpretation of a population’s age structure and helps to
understand its demographic history over the last century.
1
Northern Institute, Charles Darwin University, Australia. E-Mail: [email protected].au.
Wilson: Visualising the demographic factors which shape population age structure
868 http://www.demographic-research.org
1. Introduction
The population pyramid is one of the most popular visual representations of data in
demography. In its standard form it comprises two histograms rolled on their sides and
placed back-to-back, with the youngest ages at the bottom of the diagram and the oldest
at the top. Generally, the male population is placed on the left and the female
population on the right. The populations of each age-sex group are shown either as
absolute numbers or as a percentage of the whole population. As an example, Figure 1
presents a population pyramid for Australia at 31 December 2014.
Figure 1: The age-sex structure of Australia’s population at 31 December 2014
Source: Calculated from ABS data used in Wilson and Terblanche (forthcoming)
Demographic Research: Volume 35, Article 29
http://www.demographic-research.org 869
A population pyramid can be created in common spreadsheet or statistical software
without much difficulty, and is easy to understand. It presents immediately digestible
information on the age-sex structure of a population that would be less obvious in
tabular form. This distribution is important for understanding the demand for the wide
range of goods and services which vary by age and sex, such as baby products,
childcare, education, housing for first home buyers, household appliances, recreational
activities, aged care, and funeral services (Siegel 2002). Population age structure also
affects government spending and taxation receipts (Australian Government 2015). A
population pyramid additionally offers clues about a population’s fertility history,
mortality and migration, position in the demographic transition, and the likely influence
of its age-sex structure on future demographic change. The shape of a population
pyramid can hint at the economic or demographic role of the population in question, as
might be apparent for a university town, a popular seaside retirement region, or an area
with a communal establishment like a boarding school or prison. It can also be used to
check for potential data problems, such as age heaping. Finally, population pyramids
can also be quite entertaining, especially in dynamic form, as presented on some
national statistical offices’ and researchers’ websites. Examples can be found
• for Germany,
• for Australia, and
• for Moldova.
Several variations on the standard pyramid have been created. Many involve
disaggregating the bars into different categories of population, such as by marital status
(e.g., Statistics Canada 2013), household living arrangement (e.g., van Imhoff and
Keilman 1991), birthplace (e.g., Statistics Sweden 2009), educational attainment (e.g.,
Lutz, Cuaresma, and Sanderson 2008), labour force status (Statistics New Zealand
2015), ethnicity (e.g., Coleman 2010), and migrant status (e.g., Price 1998). Some
display shadows on the bars of the pyramid where there is an excess number of one sex
over another (e.g., Heenan 1965); others show age-sex structure with persons by age in
one histogram and age-specific sex ratios in another (Haak 1942). Commonly, two or
more population pyramids are overlaid on one another, often to compare age structure
over time (e.g., The Economist 2011), to show the outcome of alternative future
scenarios (e.g., Statistics Canada 2015), or to display probabilistic prediction intervals
(e.g., Keilman, Pham, and Hetland 2002).
However, population pyramids are not without their limitations. Because they
display only the size of each age-sex group in the population, it is usually not possible
to distinguish the relative contributions of different demographic processes to
population age structure. For example, if there is an indentation at the young adult ages,
Wilson: Visualising the demographic factors which shape population age structure
870 http://www.demographic-research.org
is this more the result of a fertility decline two to three decades earlier or recent net out-
migration? If age-specific populations at the upper end of the pyramid decline very
rapidly with increasing age, is this just the result of mortality, or have past trends in
births and migration also contributed?
This paper presents a type of population pyramid that illustrates how births, deaths,
and net migration have shaped a population’s age structure. Termed the components-of-
change pyramid, it is described and illustrated in section 2 following this introduction.
Section 3 then presents a number of examples from different countries and regions, and
demonstrates how the components-of-change pyramid helps with understanding a
population’s demographic history and why its age structure takes the shape that it does.
Section 4 consists of a discussion and conclusions, and includes a consideration of the
strengths and weaknesses of the components-of-change pyramid and potential further
developments. An Excel workbook containing the example components-of-change
pyramids accompanies this paper.
2. The components-of-change population pyramid
The size of any age group shown in a standard population pyramid equals the original
number of births of the cohort, minus the number of deaths which have occurred
between birth and the reference date of the pyramid, plus the net balance of inward and
outward migration experienced over the same age-time space. This is a simple lifetime
cohort accounting (or balancing) equation. Figure 2 illustrates these demographic
components in a Lexis diagram.
The components-of-change population pyramid displays these components in
order to aid understanding of a population’s age structure. The right-hand side of this
pyramid is similar to the standard version in that it consists of a rotated histogram of the
population by age group. The major difference is that it shows persons rather than
males and females. The left-hand side of the pyramid displays the cohort births, deaths,
and net migration volumes which produced the size of each age group. The reason for
showing population on both sides is to provide both an immediate visual representation
of age structure (on the right), and a description of its demographic drivers (on the left).
Demographic Research: Volume 35, Article 29
http://www.demographic-research.org 871
Figure 2: The cohort demographic components of change influencing the size
of age-specific populations in a population pyramid
Figure 3 shows how each of the components is displayed in the components-of-
change population pyramid. Births are depicted with a red border, deaths by grey
shading, net migration by blue stripes if positive and black stripes if negative, while
population is shown by a thick black border. These four elements are overlain in the
components-of-change pyramid itself. Part (a) of Figure 3 illustrates a case of net
migration gains to the cohort where the blue stripes for migration are shown to the left
of the bar with the red outline representing births; part (b) shows a case of net migration
losses where migration is to the left of the bar with the thick black border depicting
cohort population size.
Wilson: Visualising the demographic factors which shape population age structure
872 http://www.demographic-research.org
Figure 3: Representation of demographic components in the components-of-
change population pyramid
(a) Net migration gains
(b) Net migration losses
Demographic Research: Volume 35, Article 29
http://www.demographic-research.org 873
Figure 4 shows a components-of-change population pyramid for Australia at 31
December 2014. The red bars on the left illustrate the annual number of births, starting
with the cohort born in 1909 at the top and extending down to the cohort of 2014.
Although Total Fertility Rates are now lower than they were during the baby boom
between the late 1940s and the early 1970s, the high and sustained growth of
Australia’s population and slightly higher fertility rates over the last decade (ABS 2015)
means that recent birth cohorts have been the largest ever recorded. As the pyramid
shows, nearly all cohorts have been substantially augmented by net migration gains.
The exceptions are the youngest childhood cohorts, because these cohorts have only
accumulated a few years’ worth of net migration gains, and in Australia net migration
gains are relatively small at the youngest ages (ABS 2016a). Not surprisingly, deaths
become increasingly important in shaping age structure towards the top of the pyramid
where mortality rates are high.
Note that the final age group is age 105, not 105+. If an open-ended age group
were used it would refer to an open-ended set of cohorts whose cohort demographic
components would be hard to display in the graph. It does mean, however, that a few
very high age groups are omitted from the pyramid, but the populations at these ages
are so small they would be invisible anyway.
A key strength of the components-of-change pyramid is that it reveals the relative
importance of each of the demographic components in shaping population age structure.
For example, the bulge in Australia’s population age structure between the ages of
about 20 and 35 is not due to a mini baby boom two to three decades earlier but is the
result of substantial net migration gains. By contrast, the small peak in the age structure
centred around age 43 is due to a spike in births in the early 1970s, though the absolute
size of these age groups has been significantly augmented by net migration gains. The
minor protrusion in the age structure at ages 67 and 68 is also due to slightly more
births in 1946 and 1947 than in earlier years. These years marked the start of the post-
war baby boom. As one moves up the pyramid to the highest ages the pattern of annual
births is reflected less and less in the population age structure as mortality increasingly
becomes the dominant force.
Wilson: Visualising the demographic factors which shape population age structure
874 http://www.demographic-research.org
Figure 4: The components-of-change population pyramid for Australia,
31 December 2014
Source: Calculated from ABS data and data in Wilson and Terblanche (forthcoming)
3. More examples
Seven more components-of-change population pyramids are presented in this section as
examples. They were chosen to show a variety of age structures with differing amounts
of net migration impacts. Five use population estimates and deaths data obtained from
the Human Mortality Database (HMD 2016) with cohort net migration values
Demographic Research: Volume 35, Article 29
http://www.demographic-research.org 875
calculated as residuals. The final two examples are for subnational areas of Australia:
the state of Queensland and the Northern Territory. Their pyramids incorporate deaths
data obtained from the Australian Bureau of Statistics (ABS) and the Australian
Demographic Databank (Smith 2007), population estimates from the ABS up to age 85
and alternative estimates based on survivor ratio and extinct cohort methods at higher
ages (Terblanche and Wilson 2015), and net migration calculated as residuals. All
example pyramids show populations at 31 December for the latest year available (at the
time of writing) in each of the data sources.
3.1 England and Wales
The population age structure of England and Wales is shaped by the patterns of births
and deaths over many age ranges, but not between the mid-teenage years and the mid-
30s (Figure 5). At these ages, substantial net migration gains have augmented the size
of cohorts such that the pattern of births during the 1980s and 1990s is largely
obscured. Across other ages in the lower half of the pyramid, however, net migration
makes only a modest contribution, and it is largely the increases and decreases of births
from year to year that shape the population age profile. A dip in births around the turn
of the century, and also to a lesser extent in the mid-1970s, reveal themselves as
indentations in the age profile centred around ages 12 and 36. Similarly, the small
protrusion around age 66 is due to the spike in births immediately following the end of
World War II. The subsequent fall in birth numbers and then the baby boom of the
1950s and 1960 also shape the population profile over ages 45 to 65. Not surprisingly,
mortality is the dominant force affecting the population age structure at the oldest ages.
3.2 Scotland
Scotland’s population age structure has been shaped by migration more than that of
England and Wales (Figure 6). Some influence of net migration gains can be seen over
ages stretching from the late teenage years to the early 30s, creating a minor protrusion
centred around age 22 that would otherwise be absent. But from around age 40 upwards
the impact of migration is dramatic: huge net migration losses have depleted the
populations of all cohorts born prior to the early 1970s. Without these net migration
losses, Scotland’s population would probably be older than it is today (though in
Scotland lower net migration losses would also have led to larger numbers of births
from these cohorts). The past trend in births is also important: it has created several
indentations and protrusions in the population pyramid. Like England and Wales, there
Wilson: Visualising the demographic factors which shape population age structure
876 http://www.demographic-research.org
were declines in birth numbers around the turn of the century and in the mid-1970s and
increases in the late 1940s and during the baby boom period, which show up in the age
structure.
3.3 Finland
In contrast to the previous two examples, Finland’s population age structure shows
relatively modest net migration effects, with small net migration gains evident in the
younger adult ages and relatively minor net migration losses at older ages as well
(Figure 7). Given the concentration of migrants at the young adult ages, this is
consistent with annual migration statistics that record more emigration than
immigration until the early 1980s and then the reverse situation in more recent decades
(Statistics Finland 2007). But mostly the age structure is created by fertility and
mortality. In the bottom half of the pyramid the population age structure closely
matches the trend in annual numbers of births. In the top half of the pyramid, births still
influence the changes in population size from one age group to another, though
mortality of course becomes increasingly important in depleting populations as they
reach the highest ages.
3.4 Sweden
Sweden’s population has been augmented by substantial net migration gains in almost
every cohort (Figure 8). Official statistics show a long-standing trend of immigration
exceeding emigration, especially in recent years (Statistics Sweden 2016). Net
migration is more evenly spread across ages than in many other countries, so that over
the bottom two-thirds of the pyramid the country’s ‘roller coaster’ trend in births is the
primary cause of peaks and troughs in the age structure. The protrusions centred on
ages 23, 48, and 67 are all due to greater numbers of births in the early 1990s, mid-
1960s, and 1940s. Largely, this is due to fertility rate fluctuations. Hoem (2005)
attributes Sweden’s undulating fertility trends of the 1980s and 1990s to a mixture of
economic conditions affecting the affordability of having more children, and changes to
family benefits that provided financial incentives to have children in fairly quick
succession.
Demographic Research: Volume 35, Article 29
http://www.demographic-research.org 877
3.5 Switzerland
Switzerland’s ‘peaked’ population age profile (Figure 9) provides an interesting
contrast with the others shown in this paper. The country has clearly experienced
considerable net migration gains which have substantially augmented all cohorts in the
main working ages. The large protrusion in the age structure centred around age 47 is
due partly to the rise in birth numbers in the 1960s, but is also partly the result of larger
net migration gains than in the youngest and oldest cohorts. The cohorts aged between
their early 20s and late 30s have been increased to the greatest extent by net migration,
and obscure the lower number of births in the 1970s and 1980s. Only in the childhood
ages does the age structure closely mirror the trend in births.
3.6 Queensland
The population of Queensland has clearly been augmented by net migration gains (net
internal and net international migration combined in this case) to a very large extent
(Figure 10). The bulge in age structure between ages 20 and 35 is due to large net
migration gains rather than large numbers of births in the 1980s and 1990s. At higher
ages, however, similar amounts of net migration gains across cohorts allow the annual
pattern of births to be revealed in the age structure, with the peak of births in the early
1970s and early 1960s clearly evident. Across the childhood ages the similarity in age
group sizes is not due to a flat trend in births, but to a mixture of births and net
migration gains which varies across cohorts. At the youngest childhood ages, cohort
sizes are mostly the result of births, while at older childhood ages the cohorts started out
much smaller but have accumulated significant net migration gains. The recent increase
in the number of births in Queensland is the result of both a rise in the fertility rate and
larger populations in the childbearing age groups. At the same time, Queensland’s net
migration gains have been smaller in the last few years (ABS 2016b).
3.7 Northern Territory
The age structure of the Northern Territory has largely been shaped by migration, as
Figure 11 demonstrates. The Northern Territory is a demographically small (but
geographically enormous) remote region of Australia, subject to high rates of inward
and outward migration. Population sizes in the childhood ages roughly follow the
pattern of births over the last one and a half decades, though with fewer people in the
cohorts than at birth due to net migration losses. At all ages above 20 the population age
Wilson: Visualising the demographic factors which shape population age structure
878 http://www.demographic-research.org
structure has been increased dramatically by net (internal plus international) migration
gains, almost completely obscuring the trend in births. It can be seen that the protrusion
in the young adult ages is entirely a creation of migration, and not of a past baby boom.
Cohorts born prior to the late 1960s were increased substantially by net migration gains,
hinting at large net migration gains for the Northern Territory for many decades up to
about the late 1980s (given that peak migration is generally in the young adult ages).
Migration data indicates this to be mostly correct, with large net migration gains
recorded from the mid-1960s to mid-1980s (ABS 2014), with the exception of 1974
when a cyclone destroyed much of the Northern Territory’s capital, Darwin. Since the
mid-1980s the Northern Territory’s net migration has fluctuated between net gains and
net losses, and this is reflected in relatively smaller (but still quite large) net migration
additions to cohorts born from the late 1960s onwards.
3.8 Quantifying the effect of net migration on age structure
The example population pyramids displayed in this section present a visual
representation of the role of births, deaths, and net migration in shaping population age
structure. Given the large variations in the amounts and cohort patterns of net
migration, it is useful to quantify the influence of this demographic component in a
simple summary statistic. One approach is to define an index which describes the extent
to which cohort net migration has altered the age structure from the original numbers of
births. A Net Migration Ratio (NMR) is defined here as the absolute amount of net
migration experienced by all cohorts up to the date of the population pyramid divided
by the total original number of births, i.e.,
=
|
|
where refers to net migration, to cohort, and to births. Graphically, it is the area
on the population pyramid shaded as net migration divided by the area covered by the
red bars representing births. Figure 12 presents Net Migration Ratios for the example
populations featured in this paper.
Note that this Net Migration Ratio is not an occurrence/exposure rate or
probability; it is just a simple ratio of net migration in the cohort population accounts to
births. It is quite basic in that different patterns of net migration across cohorts could
give the same ratio, but it does have the advantage of being easy to calculate and
interpret. It cannot be given a reference year or period, as a rate or probability would,
Demographic Research: Volume 35, Article 29
http://www.demographic-research.org 879
because the net migration covers a triangular Lexis space. The time labels shown in
Figure 12 refer to the year of the population pyramid.
Figure 5: The components-of-change population pyramid for England and
Wales, 31 December 2013
Source: Calculated from data in the Human Mortality Database
Wilson: Visualising the demographic factors which shape population age structure
880 http://www.demographic-research.org
Figure 6: The components-of-change population pyramid for Scotland,
31 December 2013
Source: Calculated from data in the Human Mortality Database
Wilson: Visualising the demographic factors which shape population age structure
882 http://www.demographic-research.org
Figure 8: The components-of-change population pyramid for Sweden,
31 December 2014
Source: Calculated from data in the Human Mortality Database
Wilson: Visualising the demographic factors which shape population age structure
884 http://www.demographic-research.org
Figure 10: The components-of-change population pyramid for Queensland,
31 December 2014
Source: Calculated from ABS and Australian Demographic Databank data
Wilson: Visualising the demographic factors which shape population age structure
886 http://www.demographic-research.org
Figure 12: Net migration ratios for the case study populations in this paper
4. Conclusions
This paper presents a variation on the traditional population pyramid which offers a
visual description of the demographic factors which have shaped population age
structure. This alternative components-of-change pyramid will hopefully play a useful
part in studies seeking to understand the processes which have determined a
population’s age structure, as well as being a visualisation tool in its own right. The
pyramid is easy to comprehend and, with the Excel template accompanying this paper,
is also easy to create. Importantly, it possesses the advantage of depicting the extent to
which cohort births, deaths, and net migration are responsible for the size of each age
group. The exact contributions of these components to each age group are uncertain in a
traditional population pyramid, and ‘reading’ it without additional information can only
reveal a limited amount. To aid comparison of age structures over time and between
populations, a simple index to quantify the extent to which net migration has influenced
age structure, the Net Migration Ratio, is also suggested.
The components-of-change pyramid also has some limitations. In terms of data, it
requires births and deaths by cohort for just over 100 years, which are not available in
all countries. If net migration is calculated as an accounting residual it will incorporate
errors from all the other terms in the equation. The information the pyramid shows for
Demographic Research: Volume 35, Article 29
http://www.demographic-research.org 887
each cohort is also limited. Cohort deaths and net migration in the pyramid are
aggregated over the whole lifetime of each cohort up to the reference date of the
pyramid, and do not contain information on timing (Figure 2). However, most cohort
deaths are generally accumulated in the older ages and net migration in the younger
adult ages. In addition, there is no information on whether the presented net migration
values are the result of large or small volumes of inward and outward migration. There
is also no information by gender in the pyramids shown here.
It may also be useful to point out a potential misconception. The population age
structure which would have eventuated in the absence of migration cannot be visualised
by just ignoring the striped net migration parts of the pyramid. Migration, of course,
affects the numbers of births and deaths in a population. A population which has
experienced net migration gains will have more births and deaths because migrants
have children and die. Similarly, net migration losses reduce the numbers of births and
deaths. The population which would have resulted in the absence of migration from a
particular year can be determined by running a cohort-component projection from the
year in question that excludes migration.
Variations on the components-of-change pyramid could be constructed. For
countries and regions without the necessary cohort deaths data, it would still be possible
to plot births and then a residual deaths-plus-net-migration category. It would be less
informative, but would still indicate differences between the annual pattern of births
over the last century and the current population age structure. Other variations of the
components-of-change pyramid could show more information. If there were significant
gender differences in age structure, separate male and female pyramids could be
created. The net migration component could be broken down into different categories.
Net international migration could be split into different types, such as net migration by
broad global region, or by ‘permanent’ or ‘temporary’, providing that the requisite data
were available. At the subnational scale, net migration could be separated into net
internal and net international migration where this data exists. With sufficient time
series of data, dynamic pyramids could also be created to illustrate the changes in
demographic components shaping age structure over past decades and, using
projections, into the future as well.
Wilson: Visualising the demographic factors which shape population age structure
888 http://www.demographic-research.org
References
Australian Bureau of Statistics (ABS) (2014). Australian historical population statistics.
Catalogue no. 3105.0.65.001. Canberra: ABS.
Australian Bureau of Statistics (ABS) (2015). Births, Australia, 2014. Catalogue no.
3301.0. Canberra: ABS.
Australian Bureau of Statistics (ABS) (2016a). Migration, Australia, 2014‒15.
Catalogue no. 3412.0. Canberra: ABS.
Australian Bureau of Statistics (ABS) (2016b). Australian demographic statistics.
September 2015. Catalogue no. 3101.0. Canberra: ABS.
Australian Government (2015). 2015 Intergenerational Report: Australia in 2055.
Canberra: Commonwealth Treasury. http://www.treasury.gov.au/
PublicationsAndMedia/Publications/2015/2015-Intergenerational-Report.
Coleman, D. (2010). Projections of the ethnic minority populations of the United
Kingdom 2006‒2056. Population and Development Review 36(3): 441‒486.
doi:10.1111/j.1728-4457.2010.00342.x.
The Economist (2011). The world in 2100 [electronic resource]. London: The
Economist Newspaper Limited. http://www.economist.com/blogs/dailychart/
2011/05/world_population.
Haak, L.A. (1942). A new method of analysing the age and sex composition of a
population. Proceedings of the Oklahoma Academy of Science 23: 84‒86.
http://digital.library.okstate.edu/OAS/oas_pdf/v23/p84_86.pdf.
Heenan, L.D.B. (1965). The population pyramid: A versatile research technique. The
Professional Geographer 17(2): 18‒21. doi:10.1111/j.0033-0124.1965.00018.x.
Hoem, J.M. (2005). Why does Sweden have such high fertility? Demographic Research
13(22): 559‒572. doi:10.4054/DemRes.2005.13.22.
Human Mortality Database (HMD) (2016). Deaths, births and population data
[electronic resource]. University of California, Berkeley (USA) and Max Planck
Institute for Demographic Research (Germany). www.mortality.org.
Keilman, N., Pham, D.Q., and Hetland, A. (2002). Why population forecasts should be
probabilistic ‒ illustrated by the case of Norway. Demographic Research 6(15):
409–454. doi:10.4054/DemRes.2002.6.15.
Demographic Research: Volume 35, Article 29
http://www.demographic-research.org 889
Lutz, W., Cuaresma, J.C., and Sanderson, W. (2008). The demography of educational
attainment and economic growth. Science 319: 1047‒1048.
doi:10.1126/science.1151753.
Price, C.A. (1998). Post-war immigration: 1947‒98. Journal of the Australian
Population Association 15(2): 115‒129. doi:10.1007/BF03029395.
Siegel, J.S. (2002). Applied demography. San Diego: Academic Press.
Smith, L. (2007). Australian Demographic Databank, 1901‒2003: Deaths [electronic
resource]. Canberra: Australian Data Archive. ADA ID au.edu.anu.ada.ddi.
01118-deaths. https://www.ada.edu.au/social-science/01118-deaths.
Statistics Canada (2013). A new method of analysing the age and sex composition of a
population: Marital status: Overview, 2011. Catalogue no. 91-209-X. Ottawa:
Statistics Canada.
Statistics Canada (2015). Population projections for Canada (2013 to 2063), provinces
and territories (2013 to 2038). Catalogue no. 91-520-X. Ottawa: Statistics
Canada.
Statistics Finland (2007). Population development in independent Finland: Greying
baby boomers [electronic resource]. Helsinki: Statistics Finland.
http://www.stat.fi/tup/suomi90/joulukuu_en.html.
Statistics New Zealand (2015). National labour force projections: 2015(base)–2068
[electronic resource]. Christchurch: Statistics New Zealand. http://www.stats.
govt.nz/browse_for_stats/population/estimates_and_projections/NationalLabour
ForceProjections_HOTP15-68.aspx.
Statistics Sweden (2009). The future population of Sweden 2009‒2060. Stockholm:
Statistics Sweden.
Statistics Sweden (2016). Immigration and emigration 1960−2015 and forecast 2016‒
2060 [electronic resource]. Stockholm: Statistics Sweden. http://www.scb.se/
en_/Finding-statistics/Statistics-by-subject-area/Population/Population-
projectionPopulation-projections/Aktuell-Pong/14505/Current-forecast/The-
future-population-of-Sweden-20152060/91832/.
Terblanche, W. and Wilson, T. (2015). Accuracy of nearly-extinct-cohort methods for
estimating very elderly sub-national populations. International Journal of
Population Research 2015, article ID 978186. doi:10.1155/2015/978
186.
Wilson: Visualising the demographic factors which shape population age structure
890 http://www.demographic-research.org
van Imhoff, E. and Keilman, N. (1991). LIPRO 2.0: An application of a dynamic
demographic projection model to household structure in the Netherlands.
Amsterdam: Swets & Zeitlinger.
Wilson, T. and Terblanche, W. (forthcoming 2017). The growth of Australia’s very
elderly population: Past estimates and probabilistic forecasts. In: Swanson, D.
(ed.). The frontiers of applied demography. Dordrecht: Springer.
