2021 Census update of the Net Interstate Migration model

Methodology and results of the five-yearly review of the interstate migration model used in calculating estimated resident population.

Cite
Print
Released
15/06/2023
Source
National, state and territory population, December 2022

Overview

Net Interstate Migration (NIM) is an important component of state-level estimated resident population (ERP) change in Australia. 

Unlike natural increase and overseas migration, there is no direct measure of interstate migration. We model interstate migration using:

  • Change of address data from Medicare
  • Movements derived from Department of Defence data
  • Census-based expansion factors

Following the 2021 Census, we have:

  • recalculated the expansion factors used to account for under-reporting of address change in the Medicare data
  • used these new expansion factors to finalise quarterly interstate migration estimates for the first four years of the intercensal period (September 2016 to June 2020)
  • finalised quarterly interstate migration estimates for the last year of the intercensal period (September 2020 to June 2021) using adjusted 2021 Census data
  • applied new factors to all quarters from the 2021 Census (i.e. September 2021 onwards) and we will continue to use these until after the 2026 Census

This paper presents the outcomes of these four processes following the 2021 Census, and it includes the difference in the revision method used this Census cycle compared to previous revisions.

What are expansion factors?

Expansion factors account for missing movements due to delayed Medicare registrations and address updates, and for some usual residents not eligible for Medicare. They are multiplied by Medicare movement data to get estimates of interstate migration by state, age and sex.

Expansion factors

For the 2021 final rebasing period, expansion factors were recalculated and used to finalise net interstate migration.

    Expansion factors can:

    • account for non-reporting of moves and those not eligible for Medicare
    • expand moves based on observed misses from previous Censuses

    Expansion factors cannot:

    • react to changing trends in Medicare address reporting within an intercensal period 
    • react to changing patterns of those not eligible for Medicare (ie recent migrants)
    • take away any COVID-19 effects in our data sources

    Expansion factors:

    • are calculated for all ages but are only applied to certain age groups 
    • are calculated and applied at the state level - only the age range is determined at the Australia level

    More information about expansion factors can be found in Results - calculating expansion factors.

    COVID-19 impacts

    The COVID-19 pandemic posed some challenges to the planned approach to finalising interstate migration estimates and calculating expansion factors:

    • Medicare change of address data was affected by changed address updating behaviour due to Australia's COVID-19 vaccination program.
    • Census data was affected by COVID-19 lockdown policies in some jurisdictions.

    The response to these challenges was to give greater weight to those Census and Medicare data that were unaffected by those COVID-19 impacts. More information can be found in the Results section.

    Data sources

    Interstate migration estimates are derived from:

    • Medicare change of postcode data
    • Defence force change of postcode data
    • Census state of usual residence 1 year ago data

    For 2021, we also confronted interstate data with an administrative data asset which links multiple government data sources, including tax data. It gives additional coverage of population movements not captured by Medicare or Census. An example of how administrative data was used to improve the quality of the 2021 Census can be found in this article How administrative data improved the quality of the 2021 Census.

    Medicare data

    Quarterly estimates of interstate migration are published in National, state and territory population.

    The main input to interstate estimates is Medicare change of address information supplied by Services Australia.

    The Medicare system covers:

    • Australian citizens
    • permanent residents
    • certain temporary visa holders.

    The Medicare system excludes people on international student visas or temporary work visas (482).

    Some people, particularly younger adults, do not register changes of address with Medicare or do so long after they move. This means that the Medicare data underestimates interstate migration for certain age groups.

    Comparing Census and Medicare data over time shows that the level of under-reporting in Medicare changed between 2016-2021.

    • This coincides with the uptake of COVID-19 vaccinations in 2021 across several states and territories, which changed usual Medicare reporting behaviour.
    • Young adult males are more likely to be under-represented in the Medicare address change data, and across states and territories.
    • Additionally, there was an increase in electronic Medicare claiming options at the doctor's office, online and by mobile phone app. This increased under-reporting of address change. Under-reporting was previously confined mainly to young adults aged under 30, but over time has increasingly affected older ages. 

    The below graph shows the number of people who reported in the Census that their address one year ago was interstate from their current address, divided by the number of address changes supplied to Medicare for the same year. A ratio higher than one suggests that more people reported in the Census that they had moved interstate than had reported their move to Medicare. 

    (a) Moves over one year, Census data weighted to ERP, Medicare data unadjusted.

    (b) 2021 uses the ratio derived from 2016-2020 Census-based estimates of NIM divided by Medicare of the same period. This is to reduce COVID effects and allow a fair comparison with previous Census years.

    Despite these limitations, Medicare data is the most effective source of quarterly internal migration currently available, based on timeliness and population scope. To address some of these known limitations, we also take Defence force movements and Census 1 year ago data to supplement Medicare.

    Defence force adjustments

    Australian defence force personnel have access to alternative health services and so may not use Medicare's services. To account for this, 70% of interstate movements by defence force personnel (calculated by age, sex and state/territory of arrival and departure) are added to the Medicare data. 

    It is not known what proportion of defence personnel opt to use Medicare instead of Defence's health system. The 70% factor is an estimate based on the assumption that single people are most likely to exclusively use the Defence health service whereas personnel with a partner or dependents are likely to be listed on the same Medicare card as their family members and so captured in the Medicare address change data.

    Census data

    The Australian Census includes a question on address of usual residence one year ago and address five years ago.

    • Census one year ago is used as an alternative interstate migration estimate and an input into expansion factors.
    • Census five years ago is used for validation of results.

    These Census-based estimates complement the Medicare-based estimates.

    • Census provides a one-year and five-year snapshot of interstate migration.
    • Quarterly estimates from Medicare are needed to estimate resident population on a state and sub-state level.

    There are some limitations to Census data.

    • The 2021 Census was run during COVID-lockdowns in some states and territories, which means the data reflects a unique COVID-affected pattern.
    • The Census does not capture multiple interstate movements by the same person within the one or five-year period.
    • Non-response - either where no Census form was received from an individual, or the questions are not answered. 

    Extent of non-response in Census data

    The 2021 Census had a non-response of 5.9 % to the question on place of address one year ago.

    There was little variation in the response rate between sexes or between states (other than Northern Territory, 9.8%).

    The non-response rate did vary by age, with people over age 75 more likely not to answer.

    Adjustments made to Census data

    We made three adjustments to the raw Census interstate migration counts to counter known deficiencies.

    1. We adjusted Census data to account for residents temporarily overseas and for net undercount. This is done by taking the ratio of Census count to estimated resident population (ERP) for 10 August 2021, and applying this factor at age/sex/state level to the migration data. More information on the difference between Census population estimates and ERP estimates is in Methodology used in final rebased population estimates, June 2021.                                                                                                          \(Census\, based\,migration_{a}=Raw\,census\,migration_{a}\times {ERP\,population_{a}\over Census\,population_{a}}\) where a = age/sex/state (e.g. 20 year-old males in New South Wales).
    2. We pro-rated non-responses across states by age and sex. 
    3. People aged 0-4 on Census night have no response for the 5 year-ago question.
    • For those 0–1 year-olds we estimated these based on the one year ago data for 1-year olds.
    • For 2–4-year-olds we produced an estimate based on the data for five-year-olds by using the relationship between 5-year-olds and younger ages in the Medicare data.

    The data that resulted from this adjustment process is what we call the 'Census-based' interstate migration estimates.

    Results - Calculating Expansion Factors

    Revising interstate migration with expansion factors

    We have applied the new 2021 Census-based expansion factors to all Medicare data from September 2016 to June 2020, and they will continue to be used in the preliminary interstate migration model until after the 2026 Census. September 2020 to June 2021 quarters were finalised using Census estimates, as detailed in the section Results - Finalising Interstate Migration 2016-2021.

    Prior to this revision, preliminary interstate migration (that is, from September 2016 onwards) was based on the expansion factors calculated from the 2016 Census.

    • Data up to June 2021 is now considered final.
    • Data from September 2021 onwards will be revised following the 2026 Census. 

    The Census-based interstate migration expansion factors are available in the 'Data downloads' section.

    Calculating expansion factors

    The interstate migration estimates used in calculating the quarterly ERP incorporate each of the above data sources.

    The model includes an 'expansion factor' calculated from the Census-based estimates. This is to account for under-reporting of address change to Medicare for each age/sex/state move type (arrival or departure). The formula for the model is:

    \(Interstate\,Migration = (Medicare\,Data \times Expansion\,factor)+ (Defence\,Data \times 70\% )\)

     

    Expansion factors are calculated for each age/sex/state/move type (arrival or departure) combination, and are applied to certain age groups as:

    \(Expansionfactor = {(Census\,based\,estimate-defence\,adjustment) \times Multiple\,mover\,factor \over Raw\,Medicare\,data}\)

     

    Multiple mover factor

    The biggest conceptual difference in coverage between Census data and Medicare data is their ability to capture multiple interstate moves made by the same person within the year.

    • Medicare records up to four moves per year (one per quarter).
    • Census records a maximum of one move for the year.

    To make the two data sources more comparable, Census data is inflated by the percentage of movements captured by Medicare which are not covered by Census data.

    This exercise is to cover the following scenarios:

    • If a person moves interstate (for example, from New South Wales to the Australian Capital Territory) and then in a later quarter of the same year moves to a third state (for example, from the Australian Capital Territory to Victoria), Medicare records two moves, but Census only records one (New South Wales to Victoria).
    • If a person moves interstate (for example, from New South Wales to the Australian Capital Territory) and then returns to their original state within the same year (for example, from the Australian Capital Territory to New South Wales), Medicare will record two moves, but Census will record no move at all.

    To estimate how many records had multiple interstate movements by the same person, we calculate a multiple mover factor.

    To calculate the multiple mover factor, we matched anonymous records from the quarterly Medicare data by:

    • age
    • date of birth
    • enrolment type
    • postcode.

    We then applied the percentage results to the Census data by single year of age/sex/state.

    In 2021 the multiple mover percentage was 7.5%.

    Defence adjustments

    Census estimates include defence force personnel, so an adjustment is made to remove these extra movements. This is to allow for a direct comparison with Medicare movements, which does not cover defence force movers.

    This adjustment is for: 

    • people who moved interstate but returned to their original state, and
    • the intermediate movements of a multiple interstate move (e.g. from New South Wales to the Australian Capital Territory to Victoria).

    An example to illustrate this: 

    • If a person moves interstate and then returns to their original state, Census records no move, so these movements were removed from the defence data before subtracting from Census.
    • If a person moves (for example) from New South Wales to the Australian Capital Territory and then from the Australian Capital Territory to Victoria, Census records this as New South Wales to Victoria, so the arrival to and departure from the Australian Capital Territory are missed and therefore were removed from the defence data before subtracting from Census.

    Smoothing and capping

    Smoothing and capping are both treatments which help reduce volatility in the expansion factors. This is because the expansion factors are based on a comparison of only one year of data, and are limited in their ability to represent a longer period.

    During an intercensal period, the relationship between Census and Medicare for a given age/sex/state can fluctuate as actual migration behaviour changes. 

    In the past, expansion factors use the most recent year of data as inputs to create these factors. 

    In this rebasing, an average of four years (2016-2020) was taken to reduce the impact of COVID-19 affected data inputs for the 2021 expansion factors.

    Smoothing

    All inputs, as well as the expansion factors themselves, were smoothed by taking a three-term moving average across single years of age.

    For example, the smoothed figure for age 24 is the average of the figures for ages 23, 24 and 25. Doing this:

    • reduces the impact on future estimates of random noise within the historical data.
    • addresses the fact that Census data gives age at the end of the period, rather than age at move.

    Smoothing was not applied where this would change the real pattern of the data - i.e. defence was only smoothed starting at age 19, because smoothing ages 17 and 18 distorted the real pattern.

    Capping

    • Expansion factors were capped at 2.0, as has been the practice in the past.
    • Capping the expansion factors at 2.0 limits the influence of any one age group, whose behaviour may change over time.

    Age range

    As can be seen in the graph 'Interstate moves, ratio of Census to Medicare-based, by age' above:

    • the relationship between Census and Medicare diverges as the number of ages under or over represented increased over the last 15 years
    • In 2021, 25% of all ages had expansion factors greater than 1.0, not necessarily consecutively

    The possibility of applying all expansion factors greater than 1.0 across all ages were not considered in this final rebasing period as it would assume too much about movement patterns that can change in the next five years.

    To reduce this variability and volatility in the model as much as possible, the age range is limited to address the bulk of the under-reporting at the Australia level. 

    The age ranges for 2021 final rebasing are: 

    • 16-35 years for males
    • 16-32 years for females

    According to the 2021 Census, this covers:

    • 91% of underreporting for males
    • 96% of underreporting for females

     

    Difference between preliminary and final interstate migration estimates

    The graphs and tables below show the differences between preliminary and final net interstate migration estimates.

    According to the adjustments we made to Census data, the results were:

    • 366,000 people lived in a different state between the start and end of the 2020-21 financial year.
    • The number of interstate movements made by males and females were similar despite there being more females in the population.
    • Interstate movers had a more pronounced age distribution than the general population.
    • People aged 18 to 40 made up over half (55%) of all interstate movers, despite being around one-third (32%) of the total population.

     

    (a) Interstate movers are between 30 June 2020 and 30 June 2021; ERP at 30 June 2021

    Results - Finalising Interstate Migration 2016-2021

    Difference to previous revision method

    The method for revising interstate migration estimates is reviewed and adjusted each Census cycle in order to produce the best final estimates.

    Two key data sources for interstate migration were affected by the COVID-19 pandemic:

    • Medicare data was affected by change of address updating behaviour as people updated their details for Australia's COVID-19 vaccination program
    • Place of usual residence one year ago data from the Census was affected by COVID-19 lockdown policies

    The response to these data challenges: 

    • We changed some data inputs for expansion factors to reduce the effect of COVID-19 affected data
    • Instead of taking the most recent year of data from the intercensal period, an average of 2016-2020 financial years for Census-based estimates of NIM, a multiple movers factor, defence and Medicare was used. Using the most recent year of data gave implausible results and could not be taken forward for the next five years
    • Census-based estimates is the first data input for expansion factors and is produced by giving weights to Census data. For this rebasing, we gave a weight of 80% to 2016 Census data and a weight of 20% to 2021 Census data. This is different to the 2016 final rebasing where an equal weight was given to the two most recent Censuses.
    • Estimates for the September 2020 to June 2021 quarters were revised using Census data instead of expansion factors. This is to keep real-world effects in a year affected by COVID-19.

    Final estimates were compared with Census, linked administrative data assets and the change in intercensal difference. Intercensal difference is used as one indicator of the accuracy of interstate migration estimates with more detail in the final rebasing paper Methodology used in final rebased population estimates, June 2021.

    Previous paper: 2016 Census Update of the Net Interstate Migration Model, 2011-2016.

    Differences between preliminary and final net interstate migration, September 2016 to June 2021 quarters

    Preliminary and final NIM during this intercensal period have some key differences.

    • To reduce COVID effects in non-COVID years of the intercensal period, we finalised the first four years of the intercensal period with expansion factors, and the last year of the intercensal period with 2021 Census data.
    • ABS now uses updated Medicare data based on individual address rather than group address, which has captured more interstate movements. See more in National, state and territory population methodology, December 2022.
    • The table below compares preliminary NIM with 2016-based expansion factors using old Medicare data with final NIM with 2021-based expansion factors using new Medicare data.
    Interstate migration and corresponding intercensal difference, September 2016 to June 2021 quarters
     2016-based (preliminary)      2021-based (final)              
    Net interstate migrationIntercensal difference(a)Net interstate migrationIntercensal difference
    New South Wales-96,60094,300-118,80067,800
    Victoria28,600102,600-8,90072,400
    Queensland121,9004,000128,70014,100
    South Australia-17,200-29,500-10,700-22,400
    Western Australia-28,700-67,500-9,800-47,900
    Tasmania7,000-26,50012,700-20,300
    Northern Territory-16,300-3,300-8,8005,200
    Australian Capital Territory1,300-21,80015,700-6,400
    Australia 52,200 62,400
    1. 2016-based intercensal difference is the difference that would have occurred if NIM had not been revised. It takes into account final rebasing of the June 2021 ERP as well as finalisation of other components of ERP. This is not the same as 'preliminary intercensal difference' published elsewhere, which relates to the difference prior to final rebasing and finalisation of components. Negative numbers mean that the unrebased ERP underestimated the population. Positive numbers means that the unrebased ERP was an overestimate.
    Difference between preliminary and final net interstate migration, September 2016 and June 2021 quarters
     Difference preliminary to final 
     ArrivalsDeparturesNet(a)
    New South Wales50,00072,000-22,000
    Victoria22,00060,000-38,000
    Queensland61,00054,0007,000
    South Australia19,00012,0007,000
    Western Australia28,0009,00019,000
    Tasmania14,0008,0006,000
    Northern Territory16,0008,0007,000
    Australian Capital Territory19,0004,00014,000

    a. Net = Arrivals - Departures. Raw numbers have been rounded to the nearest thousand and may result in some differences.

    Preliminary and revised net interstate migration(a), September 2021 to September 2022 quarters
     2016-based (preliminary)2021-based (revised)Difference
    New South Wales-51,600-47,3004,300
    Victoria-17,700-15,6002,100
    Queensland63,20055,800-7,400
    South Australia1,200500-700
    Western Australia13,20013,000-200
    Tasmania-500-600-100
    Northern Territory-3,900-3,100800
    Australian Capital Territory-4,100-2,6001,500

    a. These estimates are subject to further revision after the 2026 Census. 

    2021 Census-based interstate migration expansion factors

    2021 Census-based expansion factors have now been applied to all data from the September 2016 to June 2020 quarters, and September 2021 onwards.

    Data files
    Download xlsx [88.31 KB]