FEATURE ARTICLE 1: PRELIMINARY REBASING AND COMPONENT REVISION CYCLES OF AUSTRALIA'S POPULATION ESTIMATES
After each Census, the Australian Bureau of Statistics (ABS) uses the new information obtained to rebase the estimated resident population (ERP) of Australia and its States and Territories.
Census counts by place of usual residence from the 2006 Census of Population and Housing (2006 Census) were used to construct a new base population figure for 30 June 2006. Because this new population estimate uses the Census as its main data source, it is said to be 'based' on that Census and is referred to as a population base. Rebasing refers to the process by which the ABS uses this new base derived from the 2006 Census to update the original series of published quarterly population estimates from 30 September 2001 to 30 June 2006 (the previous intercensal period). These original series of estimates (based on the 2001 Census) are now considered to be out-of-date as they have been superseded by new more accurate estimates based on the 2006 Census. For the current status of ERP refer to the table Status of Quarterly ERP Data at paragraph 7 of the Explanatory Notes.
The previous issue - Preliminary Rebasing Cycle
In the previous issue of Australian Demographic Statistics, December quarter 2006 (cat. no. 3101.0), the ABS used the 2006 Census results to produce 'preliminary rebased' population estimates for all previous intercensal quarters, including a 'preliminary base' ERP for 30 June 2006.
This issue - Component Revision Cycle
In this issue, March quarter 2007, these population estimates have been updated due to revisions of the components of intercensal population change. The ERP as at 30 June 2006, however, remains as the 'preliminary base' and has not changed.
The December quarter 2007 issue - Final Rebasing Cycle
In the December quarter 2007 issue (scheduled for release in June 2008) these population estimates will be updated again as more data becomes available to produce 'final rebased' estimates, including a new 'final base' ERP for 30 June 2006. After this, no subsequent revisions will be made to these estimates.
The remainder of this article provides a more detailed explanation of the adjustments made using the 2006 Census results and some of the related rebasing concepts.
PRELIMINARY REBASING CYCLE
Two main series of calculations using the 2006 Census results were undertaken to produce 'preliminary rebased' population estimates. These are:-
1. constructing the 'preliminary base' ERP figure for 30 June 2006; and
2. updating the previous 19 quarterly intercensal ERP figures to 'preliminary rebased'.
1. Constructing the 'preliminary base' ERP figure for 30 June 2006
This first series of calculations was applied to construct a new preliminary ERP for 30 June 2006 as a new base population from which to then estimate quarterly ERP forward. A new base population estimate is constructed for Australia and each of the states and territories. Constructing a 'preliminary base' for 30 June 2006 from the latest Census count involved two main steps.
The first step was to calculate ERP for Census night (8 August 2006). This involved:
- retabulating the Census counts by actual location to reflect a Census count by place of usual residence by:
- adding in an estimate of those residents who were absent interstate on Census night;
- subtracting those who were visiting from Interstate or were overseas visitors to Australia on Census night;
- adjusting for net undercount using the results from the Post Enumeration Survey (PES), discussed further on in this article;
- adding in the number of Australian residents who were temporarily overseas (RTOs) on Census night using data on international travellers obtained from the Department of Immigration and Citizenship; and
- applying a range of demographic adjustments designed to resolve any statistical anomalies in the age sex composition of the derived population estimates.
Adjustment components of estimated resident population, preliminary - 30 June 2006
|Components as at 8 August 2006: |
|Census count, actual location |
|plus - Residents absent interstate |
|less - Interstate visitors |
|less - Overseas visitors |
|equals - Census count, place of usual residence |
|plus - Allowance for under-enumeration(b) |
|plus - Demographic adjustment(b) |
|plus - Residents temporarily overseas(b) |
|equals - ERP as at 8 August 2006(b) |
|Backdating components to 30 June 2006: |
|less - Births(b)(c) |
|plus - Deaths(b)(c) |
|less - Net interstate migration(b)(c) |
|less - Net overseas migration(b)(c) |
|equals - preliminary ERP 30 June 2006 |
|. . not applicable |
|- nil or rounded to zero (including null cells) |
|(a) Includes Other Territories - see paragraph 2 of the Explanatory Notes. |
|(b) These estimates are preliminary and may change in compilation of final rebased estimates. |
|(c) Component data calculated for the period 1 July to 8 August 2006. |
The first part of the table on the previous page shows how the ABS arrived at the Census night 8 August 2006 ERP number by using figures obtained from applying the adjustments listed earlier. These calculations were undertaken to ensure that population estimates for the next intercensal period (i.e. 2006-2011) are as accurate as possible.
The second main step in arriving at the 'preliminary base' ERP figure for 30 June 2006, was to backdate the resulting ERP figure for Census night (8 August 2006) to 30 June 2006 using the components of population change. This involved:
- subtracting births;
- adding deaths; and
- subtracting net interstate migration and net overseas migration.
2. Updating the intercensal ERP figures during the 'preliminary rebasing' cycle
After the first series of calculations were applied to construct a new ERP base figure for Australia and each of the states and territories at 30 June 2006, a second series of calculations was made. These calculations updated the nineteen intercensal quarters preceding June quarter 2006 (i.e. September quarter 2001 - March quarter 2006) to produce 'preliminary rebased' population estimates.
To update the previous 19 intercensal quarters it is first assumed that the difference between the two intercensal base population figures (known as the 'preliminary intercensal error') is accumulated by an approximately equal number each quarter over the intercensal period. Next, in the case of the five-year intercensal period (i.e. 20 quarters), the population at the end of the first quarter is adjusted up or down by 1/20 of the intercensal error, the second quarter is adjusted by 2/20, the third by 3/20, etc., whereas the final quarter (the new base figure) is not adjusted. Details of the intercensal error for the 2001-06 intercensal period are available further on in this article.
The main reasons for updating these 19 quarterly intercensal ERP figures is to provide a more accurate population estimate for each quarter and also to ensure that the estimates from the 2001-2006 intercensal period will be comparable with all future estimates, thus creating a consistent time series of ERP data.
COMPONENT REVISION CYCLE
The 'preliminary rebased' estimates discussed above have now been updated due to the 'component revision' cycle and are published in this issue of Australian Demographic Statistics, March quarter 2007 (cat. no. 3101.0).
The size of the 'component revision' is dependent upon the timeliness and availability of updated estimates of the components of population change for the intercensal period. During the current rebasing schedule, standard annual revisions to births, deaths and NOM were undertaken for 2005-06. This annual revision altered the 2001-based series of population estimates for 2005-06, and thus changed the previous 'preliminary intercensal error' to 'updated intercensal error'. This is discussed further on in this article.
In addition, the re-derivation of interstate migration estimates for the intercensal period was also undertaken, using information from the 2006 Census on interstate movements one-year ago and five-years ago. An adjustment was then made to the 'updated intercensal error' mentioned above to take account of the use of this new Census data. Once any Census data has been used to revise a component of population growth then an adjustment is made to update the current 'intercensal error' which is then known as an 'intercensal discrepancy'. This is also discussed further on in this article.
For this 'component revision' cycle based on the 2006 Census there was only one main series of calculations undertaken once the 'preliminary intercensal discrepancy' was established. It is much the same calculation as used in 'preliminary rebasing' to update the previous 19 quarterly intercensal ERP figures.
There is no need to reconstruct another base population figure at this stage as the ERP for 30 June 2006 remains as a 'preliminary base' until the 'final rebasing'.
Updating the intercensal ERP figures during the 'component revision' cycle
During the 'component revision' cycle population data is revised for the previous 19 intercensal quarters due to the new 'preliminary intercensal discrepancy' mentioned above. Once again, as in the 'preliminary rebasing' of the intercensal ERP figures, the population at the end of the first quarter is adjusted by 1/20 of the 'preliminary intercensal discrepancy', the second quarter is adjusted by 2/20, and so on until the full previous 19 quarters have been adjusted.
The status of these 19 quarterly intercensal ERP figures plus the preliminary ERP base figure for 30 June 2006 will change to 'final rebased' in the December 2007 issue of this publication (scheduled to be released in June 2008).
INTERCENSAL ERROR AND INTERCENSAL DISCREPANCY
The intercensal error refers to the difference between the latest 2006 Census based ERP figures for 30 June 2006 and the ERP figures based on the 2001 Census which have been carried forward using births, deaths and migration data.
There are two areas that contribute to the intercensal error:
- errors in the Census based estimates of the population at the 2001 or 2006 Census date; and/or
- errors in the estimates of any of the components of population change (births, deaths and migration) during the intercensal period.
During the rebasing process the components of population change are updated as more complete data becomes available. Therefore, 'intercensal error' can be revised a few times during the rebasing process as seen in the table on the following page.
During the 'component revision' cycle information collected in the 2006 Census has allowed the ABS to estimate approximately how much of the intercensal error is due to inaccuracies in modelled estimates of interstate migration. In order to do this, the ABS has analysed data from the Census questions concerning an individual's place of usual residence one year ago, and five years ago and at Census night.
After the intercensal error is adjusted for revisions to the components of population change that use Census data such as the interstate migration revisions, the remaining unattributable portion is referred to as the intercensal discrepancy. The estimate of intercensal discrepancy for each state and territory, birth cohort and sex are spread evenly across the intercensal quarters. Thus the intercensal discrepancy acts as a balancing item, that when combined with births, deaths and migration equals the difference between the two Census population estimates. Intercensal discrepancy is caused by errors in the start and/or finish population estimates and/or in estimates of births, deaths or migration in the intervening period which cannot be attributed to a particular source. The current 2001-2006 'preliminary intercensal discrepancy' can be seen in the table below.
The table below shows the changes from 'preliminary intercensal error' to 'updated intercensal error' to the most current 'preliminary intercensal discrepancy' by number and percentage of total population for the 2001-2006 period. Final rebasing will be undertaken in 2008 and estimates published in Australian Demographic Statistics, December Quarter 2007 (cat. no. 3101.0)
Intercensal Error and Discrepancy (a), 2001 - 2006
Preliminary Rebasing Cycle
Component Revision Cycle
Final Rebasing Cycle(b)
Preliminary Intercensal Error
Updated Intercensal Error
Preliminary Intercensal Discrepancy
Final Intercensal Error
Final Intercensal Discrepancy
|New South Wales |
|South Australia |
|Western Australia |
|Northern Territory |
|Australian Capital Territory |
|nya not yet available |
|(a) A positive number indicates that unrebased ERP as at 30 June 2006 was higher than rebased ERP. A negative number indicates it was lower than rebased ERP. |
|(b) Final rebased estimates will be published in Australian Demographic Statistics, December Quarter 2007 (cat. no. 3101.0) |
|(c) Includes Other Territories - see Paragraph 2 of the Explanatory Notes. |
ADJUSTING FOR NET UNDERCOUNT
Estimated net undercount for Australia in the 2006 Census was 549,600 persons. Net undercount is the difference between the estimate of the number of people who should have been counted in the Census and the actual Census count. This estimate is based on the PES conducted in August and September of 2006. For a category of person (based on age, sex and state of usual residence), net undercount is the resultant of Census undercount, overcount, misclassification and imputation error. Adding the net undercount of people back into the population is a crucial step in arriving at the most accurate ERP possible. For more information on measuring net undercount using the PES see Information Paper: Measuring Net Undercount in the 2006 Population Census, 2007 (cat. no. 2940.0.55.001), Census of Population and Housing - Undercount, Australia 2006 (cat. no. 2940) released 5 June 2007 and Census of Population and Housing - Details of Undercount, August 2006 (cat. no. 2940.0) released 31 August 2007.
Expanding the coverage of PES
The ABS has improved the PES results by expanding its coverage to include both remote areas of Australia and discrete Indigenous communities. This expansion was undertaken to ensure that persons living in these areas and communities had an equal chance of selection for the survey. Previously, when PES did not cover these areas and communities, the ABS made an assumption that undercount in these areas were represented by survey responses for the rest of each state and territory.
New method for defining Residents Temporarily Overseas
ABS has improved the measure of net overseas migration by expanding the Australian residence criteria from 12/12 months to 12/16 months (see Information Paper: Improved Methods for estimating Net Overseas Migration (cat. no. 3107.0.55.003) and Information Paper: Statistical Implications of Improved Methods for Estimating Net Overseas Migration, Australia, 2007 (cat.no.3107.0.55.005)). This had implications for the measurement of residents temporarily overseas (RTOs) which are included in preliminary rebased ERP. Of the Australian residents absent on Census night who return within 12 months, the great majority return within five months. The ABS now uses resident status propensities based on recently observed (2005) traveller behaviour to determine the number of RTOs by inflating the observed number to represent the full 12 month period directly following the Census.
PLANS FOR FURTHER OUTPUT AND RELATED PRODUCTS
The ABS will publish final population estimates based on the 2006 Census for September quarter 2001 to June quarter 2006, in the December quarter 2007 issue of Australian Demographic Statistics (cat. no. 3101.0) to be released on 5 June 2008. That issue will include final rebased estimates dating back to 30 September 2001. No subsequent revisions will be made to those final rebased estimates.
Preliminary rebased ERP (based on 2006 Census) for the sub state/territory level is being released in two stages in Regional Population Growth, Australia, 1996-2006 (cat. no. 3218.0). Estimates for 30 June 2006 were published on 24 July 2007. Estimates for all years from June 2002 to June 2005 will be released on 2 October 2007 in accompanying spreadsheets.
Preliminary ERPs for sub state/territory level as at 30 June 2007, will be released in March 2008 in Regional Population Growth, Australia, 2006-07 (cat. no. 3218.0). Final rebased estimates for all years to June 2006 will be published in July/August 2008.
Preliminary rebased estimates of the Australian Aboriginal and Torres Strait Islander population at 30 June 2006 are available in this issue of Australian Demographic Statistics, March Quarter 2007 (cat. no. 3101.0) and were also published in Population Distribution Aboriginal and Torres Strait Islander Australians, 2006 (cat. no. 4705.0). Final rebased estimates for 30 June 2006 will be published in July/August 2008.