|Page tools: Print Page Print All|
AN EXPLANATION OF POPULATION ESTIMATES REBASING
After the 2011 Census of Population and Housing has been processed and completed, estimation of Australia's resident population using this new data source can begin. The newly compiled estimates for 30 June of the Census year can be compared with the alternative estimates produced by updating the 2006 Census-based estimates using intercensal births, deaths and migration data. The difference between these two estimates for 30 June of a Census year is called the 'intercensal error'. Of these two estimates, the more recent Census-based estimate is assumed to be the true estimate. To overcome the break in continuity that this implies, the 19 quarters of population estimates in between the two most recent Censuses (19 estimates between 30 June 2006 and 30 June 2011, out of 20 quarters of flow data) are then revised. This complete exercise (compiling estimates at 30 June of a Census year and quarterly estimates back to September of a previous Census year) is known as 'rebasing', as the population estimates are compiled from a new base, the most recent Census. Rebasing is applied to estimates of Australia's population, those of each of the states and territories, and those of sub-state geographical areas (such as Statistical Local Areas [SLAs] under the Australian Standard Geographical Classification [ASGC] and Statistical Area 2s [SA2s] under the Australian Statistical Geographical Standard [ASGS]).
Because the population base is a 'snapshot' of the population (also referred to as a stock statistic), the components used to compile the base population estimates are different from those used to compile the quarterly series of population estimates. The components used in rebasing population estimates are listed in Table 1 below.
TABLE 1. COMPONENTS USED IN COMPILING REBASED POPULATION ESTIMATES
For more information about how these components are used, please see the feature article in Australian Demographic Statistics, December quarter 2006 (cat. no. 3101.0).
Status of Estimates - Preliminary, Revised and Final
The base population described above is the starting point which is used for compiling quarterly estimates over the post-censal period until the next Census.
The components used for the quarterly post-censal estimates, also known as components of population change, are:
For more information about how these components are used to compile quarterly population estimates, please see Population Estimates: Concepts, Sources and Methods, 2009 (cat. no. 3228.0.55.001).
Through the five-yearly cycle of compiling population estimates, the estimates move through three different statuses, depending on the quality of the available components at the time the estimate is compiled. The policy of publishing preliminary estimates which are then later subject to revision is a compromise between timely availability of estimates for use in decision making and the delays in availability of the higher quality estimates of components of population growth. This trade-off means that a user can later access more accurate data by using the revised or final ERP data. At all times in the publishing cycle, the best possible estimates that can be compiled at the time are made available.
It should also be noted that at times, some components of population estimates may be impacted by other factors. While the vast majority of births and deaths are registered promptly, a small proportion of registrations are delayed for months or even years. As a result, preliminary quarterly estimates can be an underestimate of the true number of births and deaths occurring in a reference period. Revised figures for a reference period incorporate births and deaths registrations that were received after the preliminary data collection phase as well as the estimated number of registrations that have still not been received for that reference period. During the 2006-2011 intercensal period, there were several instances where processing of birth and death registrations were temporarily delayed, followed by resolution of accumulated registration backlogs by state and territory Registries of Births, Deaths and Marriages. These instances which impact on the quality and timeliness of preliminary birth and death estimates are routinely documented in explanatory notes published in Australian Demographic Statistics (cat. no. 3101.0) and also in special articles and other explanatory material in Births, Australia (cat. no. 3301.0) and Deaths, Australia (cat. no. 3302.0).
For the NOM component, preliminary estimates are available within six months after the end of the reference quarter for the production of quarterly ERP. At that time, complete traveller histories for the 16 months following a reference quarter are not available and so preliminary NOM estimation is modelled on patterns of traveller behaviours observed in final NOM one year previously. Sixteen months following a reference period, a traveller's actual duration of stay can be calculated when data on overseas movements become available. Currently, the final NOM estimates based on the '12/16 month rule' are considered to be of high quality. For more information please see Information Paper: Improving Net Overseas Migration Estimation, Mar 2010 (cat. no. 3412.0.55.001).
The Estimated Resident Population (ERP) for a reference quarter is published six months after the reference period, at which time the estimates are given the status of 'preliminary'. Rather than updating every estimate every quarter, a revision process takes place with the publication of the March and September quarter ERP. The components of ERP (births, deaths and migration) are revised as follows: births and deaths are revised every March quarter issue, and migration (NOM) is revised every March and September quarter.
TABLE 2. REVISION SCHEDULE FOR THE COMPONENTS OF POST-CENSAL POPULATION ESTIMATES
NOTE: As a consequence of these revisions to the components of population change, the ERP becomes revised.
As stated previously, following each Census, rebasing is undertaken which places estimates of the most recent intercensal period on a new base. At this time, the estimates for 20 quarters (September of the previous Census year to June of the most recent Census year) become 'final' and there is no subsequent revision planned for the components of these estimates.
For Census year estimates on the most recent Census base, there are two releases of data: 'preliminary rebased' estimates, with national and state estimates being available in June of the year following the Census, and 'final rebased' estimates, with national and state estimates being available in June, two years following the Census. Final rebased estimates for 30 June of the Census year require revised estimates of births, deaths and NOM estimates for September quarter of the Census year, before finalising the backdating of Census night population estimates to 30 June in the Census year. Preliminary net interstate migration estimates for September quarter of the Census year are used for the backdating calculation. This is because alternative estimates will not be available until the next Census is undertaken. Further, because six quarters of post-reference period border movement data are required to compile final estimates of NOM for a reference quarter, the revised estimates of NOM for September quarter 2011 will not become available until May 2013. Accordingly, 'final rebased' estimates for the 30 June 2011 will be published in June 2013.
Intercensal error and intercensal discrepancy
There are two contributors to intercensal error:
As part of the compilation process for final rebasing of population estimates, previously published national and state/territory intercensal estimates of births, deaths and net overseas migration are reviewed using the latest available source data and information and the intercensal estimates are revised where required. In the upcoming rebasing cycle, particular attention will be placed on comparing the 'stock' statistics on overseas migrants to Australia over the last five years, and the accumulated flow statistics of NOM for 2006-2011. This is because of an improved 12/16 method for estimating NOM which was implemented for the 2006-2011 intercensal period for compiling population estimates. (For more information on the 12/16 method for estimating NOM, please see Information Paper: Statistical Implications of Improved Methods for Estimating Net Overseas Migration, Australia (cat. no. 3107.0.55.005)). Further, information collected in the Census of Population and Housing allows an assessment of how much of the intercensal error was due to inaccuracies in estimates of interstate migration, using data from the two Census questions on usual residence 1 year ago and 5 years ago.
After the intercensal error is adjusted for data revisions to the components of population change, the remaining (unattributable) portion is known as the intercensal discrepancy. Thus the intercensal discrepancy acts as a balancing item, that when combined with births, deaths and migration, equals the change in the 30 June population estimates (ERP) of two consecutive Census years.
These documents will be presented in a new window.