Regional population methodology

1 This release contains estimates of the resident population of Statistical Areas Level 2 to 4 (SA2s - SA4s), Greater Capital City Statistical Areas (GCCSAs), Local Government Areas (LGAs), Significant Urban Areas (SUAs), Remoteness Areas (RAs) and electoral divisions of Australia. These estimates, plus estimates of the components of population change for SA2s and LGAs, are provided in the Data downloads section of this issue.

2 To meet the competing demands for accuracy and timeliness, there are several versions of sub-state population estimates. Preliminary estimates as at 30 June are normally available by March of the following year, revised estimates 12 months later and rebased and final estimates after the following Census. The estimates in this issue are final for 2001 to 2016, revised for 2017 and 2018, and preliminary for 2019.

3 Age and sex breakdowns of these estimates will be released on 28 August 2020 in Regional Population by Age and Sex, Australia (cat. no. 3235.0).

4 Estimated resident population (ERP) is the official estimate of the Australian population, which links people to a place of usual residence within Australia. Usual residence within Australia refers to that address at which the person has lived or intends to live for six months or more in a given reference year. For the 30 June reference date, this refers to the calendar year around it. More detailed explanations of the concept of ERP, as adopted by the ABS for official population estimates, are contained in Population Estimates: Concepts, Sources and Methods (cat. no. 3228.0.55.001).

5 Estimated resident population is based on Census counts by place of usual residence (excluding short-term overseas visitors in Australia), with an allowance for Census net undercount, to which are added the estimated number of Australian residents temporarily overseas at the time of the Census.

6 Population estimates for Australia and the states and territories (from now on referred to as states) are updated by adding to the estimated population at the beginning of each period the components of natural increase (births minus deaths, on a usual residence basis) and net overseas migration. For state populations, interstate migration is also accounted for.

7 In Australia, the SA2 (as defined in the Australian Statistical Geography Standard (ASGS)) is the base spatial unit used to collect and disseminate statistics other than those collected from the Census. In aggregate, SA2s cover the whole of Australia without gaps or overlaps. Populations for SA2s are estimated as at 30 June each year. Population estimates for larger regions are built up from SA2-level estimates.

8 The ERP as at Census date for each SA2 is calculated based on usual residence Census counts, excluding short-term overseas visitors in Australia, with an allowance for Census net undercount and the number of residents temporarily overseas (RTOs) at the Census date. The estimates of net undercount are apportioned to SA2s based on age, sex, Indigenous status, state and territory, and broad region. The number of RTOs on Census night is estimated based on coding addresses of residence to SA2 from a sample of incoming passenger cards provided by the Department of Home Affairs. As the Census is not held on 30 June (the 2016 Census was held on 9 August), further adjustments to account for births, deaths and migration for the intervening period are made to obtain the ERP at 30 June. A procedure is then applied to avoid the release of unconfidentialised usual residence Census counts while maintaining consistency with the unconfidentialised ERP.

9 SA2 populations are updated in post-Census years (from 2016) by adding to the estimated population at the beginning of each period the components of natural increase (births minus deaths), net internal migration (moves between and within the states) and net overseas migration. In some very small areas, population change since the previous Census may be assumed to be zero in the absence of reliable component data for these areas. All estimates are scrutinised and validated by ABS analysts. Local knowledge, such as that advised by state governments (including peer reviewers) is considered and used to adjust the figures for particular SA2s. Estimates at the SA2 level are constrained so that they add to the relevant state population estimates.

10 Prior to 2016, the absence of reliable migration data at the sub-state level meant that SA2 ERP was calculated using a regression model, which modelled changes in population against indicator data between the two most recent Censuses. The indicator data sources used included dwelling approvals, Medicare enrolments and Australian Electoral Roll counts. Changes in these indicators were used to estimate changes in the population of each area since the last Census.

11 In Census years, both preliminary estimates (derived from updating ERP from the previous Census) and rebased population estimates (based on the current Census) are prepared. Differences between these two sets of estimates are referred to as intercensal differences. Rebased estimates of SA2 populations for previous intercensal years are derived by apportioning the intercensal difference across the five years, while constraining to state estimates. Rebased 2012 to 2015 estimates were generally derived by adding one-fifth of the 2016 intercensal difference to the previous estimate of the 2012 population, two-fifths to the previous estimate of the 2013 population, and so on. Intercensal difference was apportioned based on the unrebased growth rate for some areas (e.g. newly established areas).

12 The SA1 is the smallest geographic unit for the release of Census data. There are approximately 57,500 SA1s and they cover the whole of Australia without gaps or overlaps. To provide some indication of ERP below the SA2 level, the ABS prepares population estimates for SA1s. Estimates are prepared at SA1 level to be aggregated to regions such as Remoteness Areas and electoral divisions. By this means, population estimates for areas other than those provided in this product (including SA1s) may be available on request.

13 Population estimates at the SA1 level as at 30 June of a Census year are compiled by apportioning the estimate for each SA2 across the SA1s within the SA2, using Census usual residence counts. In intercensal years, the 30 June population estimates for SA2s are apportioned across SA1s by taking into account population change implied by Medicare enrolments and Australian Electoral Roll counts at the SA1 level in the years following the Census.

14 In Census years, LGA ERP is prepared by aggregating whole SA2 or SA1 level estimates where possible. Where LGAs cross SA1 boundaries, Mesh Block Census counts are used to estimate the share of the SA1 population that resides in those LGAs. In intercensal years, LGA population estimates are updated by accounting for the components of population change from 2016. The components of population change (and subsequently ERP) at the LGA level are constrained to those at the SA2 level to ensure consistency between these two geographies, based on the smallest possible regions where SA2 and LGA boundaries match in terms of the combined area containing resident population. For example, where one LGA aligns exactly with one SA2 or where a group of LGAs aligns with a group of SA2s, the components for these areas will generally match.

15 To enable the comparison of regional populations over time, historical population estimates based on consistent updated geographic boundaries are prepared, for example when boundaries of Local Government Areas change. These estimates correspond with previously released estimates (on different boundaries) where possible.

16 Components of population change at the SA2 and LGA levels are calculated by breaking down state component estimates, ensuring consistency between the state and sub-state levels of geography. For further information on state component estimation, see the Explanatory Notes of Australian Demographic Statistics (cat. no. 3101.0).

17 Natural increase is calculated using births and deaths data provided to the ABS by the state and territory Registries of Births, Deaths and Marriages. Data is coded to the ASGS based on the place of usual residence of the mother for births, and the place of usual residence of the deceased for deaths. It is then aggregated to the SA2 and LGA levels and constrained to published state estimates of births and deaths.

18 Estimates of births and deaths are prepared for financial years to correspond with the 30 June reference date for sub-state ERP. To produce timely sub-state estimates, preliminary births and deaths are prepared using year of registration as a proxy for year of occurrence, which is consistent with state-level preliminary estimates. Where revisions are made at the state level, sub-state births and deaths are revised accordingly and released in the next issue of this product.

19 Preliminary birth and death estimates are subject to fluctuations caused by lags or accumulations in the reporting of births and deaths registrations, which can be caused by:

• late notification of a birth or death event to a state or territory registry;
• delays arising from incomplete information supplied for a registration;
• procedural changes affecting the processing cycles in any of the state and territory registries; or
• resolution of issues that may arise within the ABS or registry processing systems.
   

20 Birth and death registration data contributing to preliminary estimates which are higher or lower than usual at the state level are noted in the Explanatory Notes of Australian Demographic Statistics (cat. no. 3101.0), along with any explanations provided by the relevant registry.

21 The movement of people between and within Australia's states and territories is primarily estimated using Medicare change of address information. Medicare is Australia's universal health insurance scheme and covers the vast majority of Australian residents. De-identified Medicare change of address counts are provided to the ABS by Services Australia, and aggregated to SA2 and LGA levels. The data is lagged by three months to account for the time between a person changing address and updating their details with Medicare.

22 Expansion factors are applied to the Medicare migration data, by sex and age, to account for recognised undercoverage of movers across sex and age groups in the Medicare data. These factors are calculated by calibrating Medicare migration data with migration data from the previous Census. They are calculated at the state of arrival and departure level, and are applied to the sub-state Medicare migration data.

23 For defence force personnel, who access other health services and therefore may not use Medicare, the ABS uses aggregated defence force personnel movements provided by the Department of Defence. This data reflects the time of move, and is therefore is not lagged.

24 The Medicare and defence data are combined to prepare regional internal migration estimates (RIME) at SA2 and LGA levels. Interstate moves are constrained to published estimates of interstate migration.

25 RIME was previously prepared and released in Migration, Australia (cat. no. 3412.0) for financial years up to 2015-16. Users should exercise a degree of caution when comparing these estimates with the current series of RIME, due to some significant differences in the methodologies used to prepare each. The old series of RIME (for years up to 2015-16) was prepared independently of and is not directly comparable with ERP, due to the different methods and source data used. The combination of natural increase and net migration (internal and overseas) therefore may not correspond with change in ERP over this time period. The old RIME series was also prepared using quarterly postcode-based Medicare change of address data. This postcode-based data was converted to SA2/LGA, which had implications for accuracy. Further, the use of quarterly data meant that a person could record up to four moves in a financial year. The current series of RIME uses annual change of address data, consistent with the definition of population change over a financial year reference period, and is coded directly to the ASGS, removing the need to convert data from one geographical region to another.

26 Further detail on the method used to prepare postcode-based RIME for years up to 2015-16, including the use of expansion factors and defence force movements, is available in Discussion Paper: Assessment of Methods for Developing Experimental Historical Estimates of Regional Internal Migration (cat. no. 3405.0.55.001).

27 Regional overseas migration estimates (ROME) are prepared by breaking down state/territory-level net overseas migration (NOM) arrivals and departures into sub-state areas, using information from the most recent Census. For the purposes of NOM, a person is regarded as a usual resident if they reside in Australia for 12 months or more. This 12-month period does not have to be continuous and is measured over a 16-month period. It includes all people, regardless of nationality, citizenship or legal status, who usually live in Australia, with the exception of foreign diplomatic personnel and their families.

28 ROME arrivals are estimated based on counts of people who identified in the Census that they were living overseas one year ago. This distribution is used to break down state NOM arrivals each year up until the next Census. To account for changes to the distribution of overseas arrivals within a state between Censuses (e.g. in high growth areas or inner-city areas with changing numbers of temporary migrants), adjustments may be made based on up-to-date indicator data including counts of Temporary Skill Shortage visa holders and overseas students.

29 For ROME departures, a model is used to distribute state NOM departures within each state. This model is based on a range of information from the Census, mainly the number of people who arrived in each area from overseas in the last year. More weighting is given to areas that have high SEIFA Index of Education and Occupation scores and more than 20% of their total population born overseas. Of all the models evaluated, this model was selected as it best estimated population change over the previous intercensal period. As with overseas arrivals, overseas departures may be adjusted based on additional information sources.

30 LGA estimates of ROME arrivals and departures are prepared by converting SA2 ROME arrivals and departures, using a population-weighted correspondence.

31 Preliminary ROME is prepared by breaking down preliminary NOM, which is required six months after the 30 June reference period to prepare quarterly estimates of the population of Australia and the states and territories. At that time, complete traveller histories for the 16 months following a reference quarter can not be produced. Later, when preliminary estimates of NOM are finalised at the state level, ROME estimates are revised accordingly and released in the next issue of this product.

32 The status of annual sub-state ERP data changes over time, from preliminary to revised to final, as new component data becomes available at the state level. ERP for the previous year is generally revised most years due to revisions to the component data at the state level. The table below shows the status of sub-state ERP and the components of population change: natural increase, overseas migration and internal migration.

33 The sub-state population estimates in this product are produced using Census and administrative by-product data, and are subject to some error. A degree of caution should be exercised when using the estimates, especially for areas with very small populations.

34 An indication of the accuracy of ERP can be gauged by assessing the size and direction of intercensal differences - the difference between preliminary ERP for a Census year (updated from the previous Census) and rebased ERP (based on the current Census). For Australia, the preliminary (unrebased) June 2016 ERP under-estimated the final rebased June 2016 ERP by 0.1% (24,900 people). For the states and territories, the 2016 intercensal differences ranged from -1.4% (Victoria) to +2.0% (Northern Territory).

35 Summary statistics of the absolute values of these differences can broadly indicate the accuracy of sub-state population estimates. To assess the quality of SA2-based estimates prepared using the component method, experimental estimates were prepared, updated from 2011 census-based estimates using the components of population change, and compared with final rebased 2016 estimates. The average absolute value of the intercensal differences for this series of SA2 component-based estimates (excluding areas with less than 1,000 people) was 3.4%. This was slightly lower than the average absolute value of intercensal differences for regression-based estimates over the same period, at 3.5%.

36 Average absolute intercensal differences for the 2016 experimental component-based SA2 estimates generally decreased with increasing population size; that is, SA2s with large populations recorded the smallest percentage differences while small SA2s had the largest percentage differences.

37 In recognition of the inherent inaccuracy involved in estimating population, population figures in commentary text published by the ABS are generally rounded. In the commentary for this product, figures less than 1,000 are rounded to the nearest ten, figures over 1,000 are rounded to the nearest hundred, and figures over 1 million are rounded to the nearest 10,000 or 100,000. While unrounded figures are provided in summary tables and the detailed spreadsheets, accuracy to the last digit should not be assumed. Estimates of change in population are based on unrounded numbers.

38 A procedure is applied to confidentialise sub-state ERP and components, which are also subsequently constrained so that they add to relevant state population estimates. As a result of this confidentialisation method, and forced additivity, estimates of under three people should be regarded as synthetic and only exist to ensure additivity to higher levels.

39 This publication contains data presented according to the 2016 edition of the Australian Statistical Geography Standard (ASGS), which refers to boundaries as defined at 1 July 2016. Under this classification, statistical areas are defined as follows:

• Statistical Areas Level 2 (SA2s). SA2s are medium-sized general purpose areas which aim to represent communities that interact together socially and economically. SA2s are based on officially gazetted suburbs and localities. In urban areas SA2s largely conform to one or more whole suburbs, while in rural areas they generally define the functional zone of a regional centre.
• Statistical Areas Level 3 (SA3s). SA3s are aggregations of whole SA2s and reflect a combination of widely recognised informal regions as well as administrative regions such as state government regions in rural areas and Local Government Areas in urban areas.
• Statistical Areas Level 4 (SA4s). SA4s are made up of whole SA3s and are designed to reflect labour markets. In rural areas, SA4s generally represent aggregations of small labour markets with socioeconomic connections or similar industry characteristics. Large regional city labour markets are generally defined by a single SA4. Within major metropolitan labour markets SA4s represent sub-labour markets.
• Greater Capital City Statistical Areas (GCCSAs). GCCSAs are built from whole SA4s and represent a broad socioeconomic definition of each of the eight state and territory capital cities. They contain not only the urban area of the city, but also the surrounding and non-urban areas where much of the population has strong links to the capital city, through for example, commuting to work.
• Significant Urban Areas (SUAs). SUAs are aggregations of whole SA2s which represent concentrations of urban development with populations of 10,000 people or more. They do not necessarily represent a single Urban Centre, as they can represent a cluster of related Urban Centres with a core urban population over 10,000 people. They can also include related peri-urban and satellite development and the area into which the urban development is likely to expand. SUAs may cross state/territory borders.
• Remoteness Areas (RAs). RAs represent an aggregation of non-contiguous geographical areas which share common characteristics of remoteness. The delimitation criteria for RAs are based on the Accessibility and Remoteness Index of Australia (ARIA+), which measures the remoteness of a point based on the road distance to the nearest urban centre. The RA categories range from Major Cities to Very Remote. Each RA is created from a grouping of SA1s which have a particular degree of remoteness. Data for RAs are prepared by aggregating the data for SA1s.

40 This product also contains data presented according to the 2018 edition of the Australian Statistical Geography Standard (ASGS) - Non ABS Structures:

• Commonwealth Electoral Divisions (CEDs). A CED is an area legally prescribed for returning one member to the House of Representatives, Australia's Federal Lower House of Parliament. Data for CEDs are approximated by aggregating the data for SA1s that best fit the area.

41 This product also contains data presented according to the 2019 edition of the Australian Statistical Geography Standard (ASGS) - Non ABS Structures:

• Local Government Areas (LGAs). LGAs are ABS approximations of officially gazetted local government boundaries as defined by each state and territory local government department. LGAs cover incorporated areas of Australia, which are legally designated areas for which incorporated local governing bodies have responsibility. The ABS updates LGAs annually, and prepares updated and historical population estimates based on these updated boundaries.
• State Electoral Divisions (SEDs). An SED is an area legally prescribed for returning one or more members to the State or Territory Lower Houses of Parliament. Data for SEDs are approximated by aggregating the data for SA1s that best fit the area.

42 The population of the Other Territories, namely Christmas Island, Cocos (Keeling) Islands, Jervis Bay and Norfolk Island, is included in all references to the total population of Australia. However, the Other Territories are excluded from the commentary in this product.

43 Further information on these statistical areas is contained in:

Australian Statistical Geography Standard: Volume 1 - Main Structure and Greater Capital City Statistical Areas, July 2016 (cat. no. 1270.0.55.001)

Australian Statistical Geography Standard: Volume 3 - Non ABS Structures, July 2018 (cat. no. 1270.0.55.003)

Australian Statistical Geography Standard: Volume 3 - Non ABS Structures, July 2019 (cat. no. 1270.0.55.003)

Australian Statistical Geography Standard: Volume 4 - Significant Urban Areas, Urban Centres and Localities, Sections of State, July 2016 (cat. no. 1270.0.55.004)

Australian Statistical Geography Standard: Volume 5 - Remoteness Structure, July 2016 (cat. no. 1270.0.55.005)

44 Maps for Australian statistical areas are available in the online mapping tool ABS Maps. A complete series of SA2 maps is available to download from Australian Statistical Geography Standard: Volume 1 - Main Structure and Greater Capital City Statistical Areas, July 2016 (cat. no. 1270.0.55.001).

45 Regions are ranked by both largest growth and highest growth rate. Largest growth is based on the absolute change in population between June 2018 and June 2019, while highest growth rate is based on the percentage change. Regions with less than 1,000 people at June 2018 are excluded from the highest growth rate rankings.

46 Due to the inherent imprecision of regional population estimates and variation in population size, rankings should be considered indicative of relative growth between regions, not definitive.

47 The centre of population is a measure used to describe the spatial distribution of a population. The method used to calculate centres of population in this product is based on the centroid and population of each Statistical Area Level 1 (SA1). To calculate the centre of population for an area, the latitude and longitude coordinates of the centroid of each SA1 in that area are multiplied by the SA1's ERP to obtain weighted latitudes and longitudes for each SA1. These are summed to obtain a weighted latitude and longitude coordinate for the area, then divided by the total population of the area to obtain a single latitude and longitude coordinate.

48 Due to the inherent imprecision in small area estimates, the centre of population should be considered indicative only of the distribution of population, and cannot be ascribed to an exact location. The use of different geographical level data can result in different centres of population.

49 The area figures used in this issue are based upon the SA2 level of the 2016 edition of the ASGS. The areas of the SA2s were calculated using ABS standard Geographic Information Systems software from the digital boundaries of this ASGS edition. Higher level spatial unit area figures are aggregations of the relevant SA2 areas. These areas are included in the SA2-based ERP spreadsheet accompanying this release. Area figures are also provided for LGAs based on the 2019 edition of the ASGS and can be found in the LGA-based ERP spreadsheet.

50 The population density of an area as featured in the Excel spreadsheets in this product has been calculated by dividing its ERP by its area in square kilometres. The result is expressed as a number of people per square kilometre.

51 In this release, estimated resident population data has also been published in 1km² grid format. The population grid offers a consistently sized spatial unit and gives a refined model of population distribution, particularly for the non-urban areas of Australia. It is also an established, easy to understand and readily comparable international standard which will enable users to make local, national and international comparisons of population density.

52 The population grid was modelled using preliminary 2019 SA1 ERP. All SA1s with an ERP greater than zero were identified. Within these SA1s all known residential dwelling locations were identified using a variety of sources including the Geocoded National Address File (GNAF).Within each populated SA1, the 2019 SA1 ERP was distributed equally across all the residential dwellings. The average value assigned to each dwelling was then summed within each 1km² grid cell across the country.

53 The population grid is provided in the following formats:

• ESRI Grid format which can only be opened in a Geographic Information System (GIS).
• GeoTIFF format is a Tagged Image File Format (TIFF). It is a raster graphics file format that is widely supported by graphics software. The Geo extension to the TIFF format is a metadata storage format which allows georeferencing information (datums, ellipsoid, coordinate systems, map projection) to be embedded within the TIFF file. These metadata allows Geographical Information Systems (GIS) software, such as MapInfo, ArcGIS or QGIS, to correctly interpret the location of the image and compare the image with other spatial referenced data.
• PNG format is a Portable Network Graphics File (PNG). It is a raster graphics file format that is widely supported by graphics software including those bundled with the major operating systems (Microsoft Windows, Apple OS X & iOS). The objective of publishing in PNG format is to allow users to quickly visualise a "picture" of these data.
   

54 The ABS is exploring the development of temporary population estimates as part of ongoing efforts to use existing available data to meet this information need. The ABS has also partnered with the University of Queensland to further explore the issues associated with estimating temporary populations. Obtaining better information on daytime/night-time, weekday/weekend and seasonal population numbers would give policy makers a greater understanding of these populations. Such data would better inform population-based funding decisions in areas such as health, education, transport, and in areas with fly-in, fly-out populations.

55 The ABS has undertaken a pilot project that explored aggregated telecommunications data at SA2 level to estimate hourly temporary populations. The pilot study covered three weeks throughout 2016, including the week around Census night. The project revealed some potentially promising insights, with the aggregated population data in line with expectations across geography and time. The ABS will continue to investigate the feasibility of estimating non-resident populations, subject to resources and accessibility to data sources.

56 ABS publications draw extensively on information provided freely by individuals, businesses, governments and other organisations. Their continued cooperation is very much appreciated; without it, the wide range of statistics published by the ABS would not be available. Information received by the ABS is treated in strict confidence as required by the Census and Statistics Act 1905.

57 Other ABS releases that are freely available on the ABS website and may be of interest to users of this product include:

Australian Demographic Statistics (cat. no. 3101.0)

Regional Population by Age and Sex, Australia (cat. no. 3235.0)

Births, Australia (cat. no. 3301.0)

Deaths, Australia (cat. no. 3302.0)

Migration, Australia (cat. no. 3412.0)

Australian Historical Population Statistics (cat. no. 3105.0.65.001)

Population Projections, Australia (cat. no. 3222.0)

Information Paper: Population Concepts (cat. no. 3107.0.55.006)

Population Estimates: Concepts, Sources and Methods (cat. no. 3228.0.55.001)

Quality Assurance of Rebased Population Estimates (cat. no. 3250.0.55.001)

Data by Region

58 The ABS may have other relevant data available on request. Inquiries should be made to the National Information and Referral Service on 1300 135 070. The ABS Privacy Policy outlines how the ABS will handle any personal information that you provide to us.

59 Statistics in this publication predate the Australian summer bushfires of 2019-20 and COVID-19.

There were no LGA name, code or boundary changes involving population in the 2019 edition of the ASGS.