• AIHW Data Quality Statements, Superseded 23/11/2022

Description

All states and territories provide a range of public housing programs and maintain administrative data sets about these programs. Extracts of these data sets are provided annually to the Australian Institute of Health and Welfare (AIHW). This collection contains information about public housing (PH) dwellings, households assisted and households on the waiting list. Data are provided for the following two reference periods: 2020–21 financial year and point in time at 30 June 2021.

Summary

• The data collected are an administrative by-product of the management of PH programs run by the states and territories and conform well in terms of scope, coverage and reference period.
• The administrative data sets from which this collection is drawn have inaccuracies to varying degrees, including missing data, out-of-date data and data coding or recording errors.
• Care is required when comparing outputs across states and territories. Differences in the data collected and which records are included or excluded from a calculation, can affect the coherence of the outputs. Coherence over time has also been affected by changes in methodology (see ‘coherence’ section for details).

The Australian Institute of Health and Welfare (AIHW) is an independent corporate Commonwealth entity under the Australian Institute of Health and Welfare Act 1987 (AIHW Act), governed by a management board and accountable to the Australian Parliament through the Health portfolio.

The AIHW is a nationally recognised information management agency. Its purpose is to create authoritative and accessible information and statistics that inform decisions and improve the health and welfare of all Australians.

Compliance with the confidentiality requirements in the AIHW Act, the Privacy Principles in the Privacy Act 1988 (Cth), and AIHW's data governance arrangements ensures that the AIHW is well positioned to release information for public benefit while protecting the identity of individuals and organisations.

For further information, see the AIHW website https://www.aihw.gov.au/about-us, which includes details about the AIHW's governance (https://www.aihw.gov.au/about-us/our-governance) and vision and strategic goals (https://www.aihw.gov.au/about-us/our-vision-and-strategic-goals).

The AIHW receives, compiles, edits and verifies the data in collaboration with states and territories. The finalised data sets are signed off by the states and territories and used by the AIHW for reporting, analysis and approved ad hoc data requests. Requests for jurisdiction-level data must be signed off by the relevant state or territory.

The reference period for the PH collection is based on the financial year (ending 30 June). The specific reference period for these data is 2020–21.

Analysis of data from this collection is reported in the AIHW's annual Housing assistance in Australia reports and the Productivity Commission's annual Report on government services.

Users can request data not available online or in reports (subject to the AIHW's confidentiality policy and state and territory approval) via the AIHW’s online data request system at https://www.aihw.gov.au/our-services/data-on-request. Depending on the nature of the request, access to unpublished data may incur costs or require approval from the AIHW Ethics Committee.

General enquiries about the AIHW publications can be directed to [email protected].

Metadata and definitions relating to this data source can be found in the Public Housing and State Owned and Managed Indigenous Housing (PH & SOMIH) data set specification 2018-.

Supplementary information can be found in the housing collection data manuals which are available upon request.

The data collected are an administrative by-product of the management of public housing programs run by the states and territories and conform well in terms of scope, coverage and reference period.

Classifications used for income, disability status, greatest need and vacancy reason are not consistent across the states and territories, however, the states and territories map these data to an AIHW standard as defined in METeOR.

There are known accuracy issues with the data collected:

• The administrative data sets from which this collection is drawn have inaccuracies to varying degrees including missing data, out-of-date data and data coding or recording errors.
• Not all states and territories capture all data items. For those outputs that are calculated using gross income, New South Wales and South Australia use assessable income instead. (Assessable income is the income amount used to establish eligibility for housing assistance and each jurisdiction uses its own definition of assessable income.) In addition, disability status is derived using the receipt of a disability pension as a proxy in Victoria. New South Wales, Queensland, Tasmania and the Australian Capital Territory reference payment type as well as other information.
• Information about disability is not reported under eligibility requirements in some jurisdictions.
• Indigenous status is self-identified and not reported under eligibility requirements in some jurisdictions.
• Most states and territories do not update income information for non-rebated households, that is, households who pay the market rent value of the dwellings. Therefore, some household income information may be coded as missing or not reflect current income levels. Outputs that require income information do not include households with missing income and therefore may not be complete.
• All jurisdictions have a form of integrated waitlist across social housing programs. Waitlist data are reported separately for each social housing program applied for. Counting rules for the number of applicants for any given program may vary across jurisdictions.

State- and territory-specific issues:

New South Wales

• Since a system change in 2010, New South Wales continues to report problems encountered when linking files containing date variables within their system. This may occur when linking ‘dwelling history’, ‘household’ and ‘waitlist’ files. Where date variables contradict between files, they are recoded as missing.
• Data on ‘number of Indigenous households at 30 June’ are incomplete and unreliable. These data are substituted with estimates from the 2016 census.
• Income details are only updated for rebated households, that is, 94% of all households.

Victoria

• Income details are only updated for rebated households, that is, 91% of all households.
• Victoria does not collect current rent paid for waitlist applicants and are therefore unable to determine households who are in greatest need due to very high housing costs. These factors lead to an undercount of greatest needs allocations.
• The first name and surname components used to calculate the statistical linkage key (SLK) were not provided.

Queensland

• Income details are only updated for rebated households, that is, 96% of all households.
• There is one waiting list for all social housing in Queensland. Applicants for the PH program may also be reported in the community housing waitlist data. Applicants eligible for both the PH and state owned and managed Indigenous housing (SOMIH) programs are reported only in the SOMIH waitlist data.
• Market rent is only current for occupied dwellings.

Western Australia

• Indigenous data is voluntarily recorded and known to be an undercount. However, recording of Indigenous status is more accurate for priority applications, resulting in higher counts of confirmed Indigenous households than in previous years.
• As disability is only recorded if it contributes to housing needs, many people with disability are not identified. As a result, special needs allocations will be undercounted.
• Income details are only updated for rebated households, that is, 93% of all households.
• Western Australia does not collect current rent paid for waitlist applicants and are therefore unable to determine households who are in greatest need due to very high housing costs. These factors lead to an undercount of greatest need allocations.
• In some cases, greatest need date is not recorded and is assumed to be the application date.
• Inconsistent recording of dwelling vacancy dates and vacancy reasons impacts the accuracy of turnaround time calculations.

South Australia

• Income details are not updated for most non-rebated households.
• From April 2020, eligible applicants were registered for both the PH and SOMIH programs. These applicants are reported only in the SOMIH waitlist data.
• Housing SA did not supply the ‘dwelling history’ file for 2020–21, which includes the variables necessary for the calculation of turnaround time.

Australian Capital Territory

• Income details are only updated for rebated households, that is, 94% of households.
• The first name and surname components used to calculate SLK were not provided.
• Relationship status for household members is not consistently recorded resulting in a number of unknown values for overcrowding and underutilisation measures.

States and territories may publish their own analysis of public housing data which may vary in scope from this collection.

Data for individual states and territories may not be comparable to previous years due to changes in systems and processes which have led to differences in the accuracy and completeness of the data over time. Differences between states and territories concerning social housing management systems, incomplete or missing information, out-of-date information and coding errors can affect the coherence of the outputs.

Coherence over time has also been affected by changes in methodology:

• From 2018–19, remoteness area (RA) is determined using a concordance between 2018 postcodes and the Australian Bureau of Statistics (ABS) 2016 RA classification. For 2017–18, a concordance between 2017 postcodes and the 2016 RA classification was used. Previous years used a concordance between 2012 postcodes and the 2011 RA classification. Care is therefore required when comparing remoteness data across time.
• Measurements using low income cannot be compared with low income measures produced prior to 2009–10 due to a change in methodology. From 2009–10 onwards, low income cut-offs were obtained from the biennial ABS Survey of Income and Housing (SIH). From 2018–19, low income cut-offs are based on 2017–18 SIH results. Care is required when comparing low-income measures over time.
• Measurements of overcrowding cannot be directly compared with figures produced prior to 2009–10 due to a change in methodology. Prior to 2009–10, overcrowding was measured using a proxy standard of ‘2 or more bedrooms are required’. From 2009–10 onwards, overcrowding was measured using the Canadian National Occupancy Standard (CNOS) of ‘1 or more bedrooms are required’.
• Measurements of underutilisation cannot be directly compared with figures produced prior to 2011–12 due to changes in methodology. Prior to 2009–10, underutilisation was measured using a proxy standard of ‘2 or more spare bedrooms’. From 2009–10 to 2010–11, underutilisation was measured using the CNOS of ‘1 or more spare bedrooms’. From 2011–12 onwards, underutilisation was measured using the revised CNOS of ‘2 or more spare bedrooms’.

State and territory government housing authorities’ bedroom entitlement policies may differ from the CNOS which is used in dwelling utilisation calculations.

State- and territory-specific issues:
New South Wales

• From 2011–12, previous rent charged is not a mandatory field to be provided by new tenants in their application for housing assistance. As a result, the number of new allocations to households in greatest need is underestimated due to the exclusion of households with very high rental housing costs.
• A new maintenance system was introduced during 2016–17. The transition and implementation timing meant there were significant data gaps in the ‘dwelling history’ file. Therefore, New South Wales did not supply the 2016–17 ‘dwelling history’ file which includes the variables necessary for the calculation of turnaround time.
• The 2016–17 waitlist data excluded suspended applicants, therefore data may not be comparable to previous years.
• In 2017–18, around 1,000 dwellings identified for disposal and another 1,000 dwellings leased to a community organisation were reported for the first time. Person information is not available for these dwellings.

Victoria

• In 2017–18, a system error was corrected that previously caused incorrect recording of vacancy availability. Consequently, turnaround time information for Victoria is not comparable with previous years.

Queensland

• There was a major system change in 2019–20. This affected the following:
  • Migration issues meant Queensland were unable to report household income for some households.
  • Changes to the categorisation of vacancy reason.
  • Identifiers (for example, dwelling and household ID) are not consistent with previous years meaning descriptors related to sustaining a tenancy for 12 months or more could not be calculated.
  • The methodology for differentiating between PH and State Owned and Managed Indigenous Housing (SOMIH) applications changed.
  • The methodology for collecting greatest need information changed.

Western Australia

• Housing Authority WA changed core systems during 2014–15 resulting in a number of differences in recording practices and an amalgamation of data from the old and new systems. The 2015–16 reporting period was the first full year in which the new system was the principal data source. Therefore, any comparisons with previous years for Western Australia data should be undertaken with caution.
• Collection methodology changed for 2016–17. The variation from figures used in previous years was minimal. In 2016–17, some priority households were housed from the general waitlist rather than the priority list meaning they were not reported as greatest need allocations.

South Australia

• Since 2012–13, Housing SA has been unable to provide the ‘dwelling history’ file which includes the variables necessary for the calculation of turnaround time.
• In April 2020, Housing SA implemented a new data collection system. A number of data migration errors and coding changes in the new system led to reporting errors or an inability to provide items in 2019–20. In particular:
  • Dwelling and tenancy details were unable to be provided for some PH dwellings.
  • Income unit identifier was not provided.
  • Greatest need is likely to be an undercount due to a data migration error.
  • South Australia can no longer distinguish between greatest need reasons. For greatest need households, greatest need reason is reported as homeless (code 1).

However, data cleansing before migration to the new system resulted in better data quality and less missing information regarding bedroom and/or required bedroom details.

Tasmania

• Housing Tasmania implemented a new Housing Management System (HMS) in November 2016. There was a gap in the data collection from 2016–17 when income details and wait times were not available for applicants who registered prior to October 2016. This is improving over time.
• In previous years, waitlist data included transfer applicants who were not on the housing register. From 2017–18 this is no longer the case and caution should be taken when comparing waitlist data to previous years.
• In 2018–19, relationship status for household members was not consistently recorded resulting in a number of unknown values for overcrowding and underutilisation measures.
• Greatest need reason was not reported until 2018–19.

Australian Capital Territory

• A change in the facilities management provider (for maintenance and property upgrades) occurred in 2018–19. Miscoding of vacancies has not allowed for accurate distinction between routine and non-routine maintenance which impacts the accuracy of turnaround time calculations.
• In 2018–19, system processing of rebates approved in the last week of the financial year occurred after the end of financial year data extraction. This resulted in a small (1%) undercount of households on a rebate and an apparent increase in households paying more than 30% of their income on rent.
• In 2019–20, the increase in the number of households with at least one member who reports to experience disability is likely due to an improvement in data quality.

Northern Territory

• Around 5,000 social housing dwellings in the Northern Territory were transferred from the Indigenous Community Housing collection to remote public housing between 2008 and 2010 and have been excluded from administrative data collections between 2008–09 and 2015–16, inclusive. From 2016–17, these dwellings were reported in SOMIH. Due to data quality issues, limited aggregate information was provided in 2016–17. Unit record data were provided from 2017–18.
• Prior to 2018–19, reinstated applications (where contact with a new applicant had been lost then re-established) were reported as transfers instead of new households. In 2018–19, these applications were reported as new households. As a result, in previous years total new applicants (excluding transfers), new greatest need applicants and new households assisted were understated and transfer applicants and relocating households were overstated. From 2019–20, reinstated applicants who relocated to another dwelling in the same housing program but chose to wait for their transfer in private accommodation are counted as transfers.
• In 2018–19, dwellings returning to service after receiving major upgrades under stimulus programs were a contributing factor to high turnaround times.
• In 2019–20, more households provided income details. This may reflect an increase in households who have provided consent for Centrelink income data to be collected automatically.
• In 2019–20, the decrease in total vacancy days is partly due to improved classification of vacancy reason.

Australian Institute of Health and Welfare