3228.0 - Demographic Estimates and Projections: Concepts, Sources and Methods, 1999  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 30/08/1999   
   Page tools: Print Print Page Print all pages in this productPrint All  
Contents >> Chapter 6. Population projections


6.1. ABS produces a range of population projections, both published and unpublished. ABS commenced its involvement in this area in the late 1940s and began publication in the late 1970s. Published projections relate to the total population down to capital city/balance of State level and for the Indigenous population down to the State level. In late 1999 projections by household composition will also be published. Unpublished projections using customised assumptions are also produced for populations that include the total population, the Indigenous population and the electoral roll population. Such unpublished projections are not official ABS statistics and should generally be referred to as projections produced the ABS according to assumptions agreed to by (the client).

6.2. Currently ABS publishes population projections for Australia, States and capital city / balance of state twice every five years in Population Projections (3222.0). Various scenarios of future population size, structure and distribution are presented based on differing assumptions of fertility, mortality, overseas migration and internal migration, further details of which can be found in the publication.

6.3. The projection technique is the cohort component method, essentially identical to that used for ABS' post-censal population estimates (see Chapter 2, paragraph 2.15). It involves the application of fertility, mortality and migration assumptions by age and sex to a base year population (usually the latest available population estimates) to derive a projected population for the next year. That population then has the following year's assumptions applied to it to produce a projected population a year further, and this process is repeated until the projection horizon is reached.

6.4. While the assumptions, and therefore the projection results are confined to demographic disaggregation by age, sex and geographical level, additional information on derived growth rates, birth and death levels, median age, working/school/retirement-age populations is also presented in the publication. Sections are also included detailing the background to the assumptions and analysis of the projection results.

6.5. The cohort-component method is employed for each geographical level and sex, the formulae are shown in paragraphs 6.6 to 6.12 and use the following variables:

highest age projected (currently States: 100+; Sub-state: 85+)
base year
fertility rate
death probability
net overseas migration
net interstate migration
interstate departures
interstate arrivals
per capita departure rate (donor state)
per capita arrival rate (receiving state)

6.6. Survivorship rates (1 - death rates) are applied to the population thereby 'ageing' it one year, for ages 0 to max - 1. Overseas migrants are assumed to arrive on average half-way through the projection year, so (a) they only have half a year's mortality rates applied, and (b) are half an age older than Px, therefore requiring new ages to be formed by combining two halves of consecutive ages.

Px(t) * [ 1 - Qx(t) ]
[ 0.5 * OMx(t) ] * [ 1 - ( 0.5 * Qx(t) ) ]
[ 0.5 * OMx+1(t) ] * [ 1 - ( 0.5 * Qx+1(t) ) ]

6.7. A similar process is performed for the highest age group:

Pmax(t) * [ 1 - Qmax(t) ]
Pmax-1(t) * [ 1 - Qmax-1(t) ]
OMmax(t) * [ 1 - ( 0.5 * Qmax(t) ) ]
[ 0.5 * OMmax-1(t) ] * [ 1 - ( 0.5 * Qmax-1(t) ) ]

6.8. Births are then calculated by applying calendar year fertility rates to the mid year population. As births will occur on average half-way through a projection year (ie. t+), the required births are the average of when fertility rates are applied to year t and year t+1 populations.


 0.5 * [ Fx(t) * Pf,x(t) + Fx(t+1) * Pf,x(t+1)]


6.9. Population aged zero is then calculated by 'surviving' the births (from t+ to t+1) and adding 'survived' overseas migration.

B(t) * [ 1 - Qb(t) ] + [ 0.5 * OM0(t) ] * [ 1-( 0.5 * Q0(t) ) ]

6.10. Where interstate migration is required, it is calculated by applying departure rates to a state's population (those at risk of departing) and arrival rates to the rest of Australia's population (those at risk of moving to that state). These rates are derived from 1996 Census data and are held constant for the duration of the projection.

DEPx, s(t+1)
Px, s(t+1) * D_RATEx

ARRx, s(t+1)
Px, not-s(t+1) * A_RATEx

6.11. The resulting total arrivals and departures are then scaled to the pre-determined net interstate migration assumption. Finally, the arrivals and departures by age and sex are scaled to the new arrival and departure totals (while ensuring they net across all States to give zero for Australia), then combined to give net interstate migration.

ARRx,s(t+1) - DEPx,s(t+1)

6.12. Then add the interstate migration:

Px(t+1) + IMx(t+1)

6.13. A final proration is then undertaken to ensure lower geographic levels sum to the projected age/sex population of any higher geographic level. Year t+1 then becomes the base for projecting the next year and the cycle is repeated until the final projection year is reached.

Experimental Projections of the Indigenous Population

6.14. ABS projections of the Aboriginal and Torres Strait Islander (Indigenous) population are referred to as experimental because of the experimental nature of the base population and the deficiencies in the quality of Indigenous births, deaths and migration data involved in deriving the population projection assumptions. The inclusion of an assumption for change in propensity to identify as Indigenous on a census form also adds to the experimental nature of the projections. The most recent edition is Experimental Projections of the Aboriginal and Torres Strait Islander Population, 1996 to 2006 (3231.0).

6.15. The base population for these projections is the 30 June 1996 estimate of the Indigenous population. The method of estimation is detailed in Appendix 5 and Experimental Estimates of the Aboriginal and Torres Strait Islander Population, 1991 to 1996 (3230.0).

6.16. As with projections of the population as a whole, Indigenous projections employed the cohort-component method. The formulae above therefore also apply to projecting Indigenous population, however two additional steps were included:

      • Allow for an increase in the propensity of people to identify as being of Indigenous origin; and
      • The inclusion of 'paternity rates', that is births of Indigenous children from an Indigenous father and a non-Indigenous mother.
6.17. The difference between the 1991 and 1996 Census counts of Indigenous persons, was significantly larger than expected despite similar collection procedures for the two censuses. This increase cannot be fully accounted for by natural increase and net migration over the intercensal period, but is evidence of a large increase in the propensity to indicate Indigenous origin on census forms between the two censuses, as was also the case between the 1986 and 1991 Censuses. For more information on propensity to identify as Indigenous, see Appendix 5 - Estimating the Indigenous Population.

6.18. The projected low-growth scenario assumed no future change from the 1996 propensity to identify as Indigenous. The high-growth scenario assumed that the average annual change in propensity to identify between 1991 and 1996 Censuses will continue. This adjustment to the population was made at the end of each year's projection cycle (ie. after (8) above) thus:

Px(t+1) * R
where R is the state-specific population growth factor due solely to change in propensity to identify.

6.19. Unlike in general population projections where the number of births can be expressed entirely as a function of the number of women in childbearing ages, Indigenous births may also come from Indigenous men and non-Indigenous women. These birth rates are referred to as 'paternity rates', while births to Indigenous women are, as usual, referred to as fertility rates. Assumed paternity rates are applied in the projection at the same stage as fertility rates (ie. at formula (3) above) using an equivalent formula:


0.5 * [ Mx(t) * Pm,x(t) + Mx(t+1) * Pm,x(t+1)]

where m is the male population and M is the paternity rate.

Previous PageNext Page