3.6 Dwellings

Our needs and preferences for housing change over time, and differ from region to region. Information gathered in the Census about dwellings is used to determine changes in levels of housing stock and patterns of use, including changes in housing density. It also assists in urban and neighbourhood design, transport planning, and land use forecasting.

The 2021 Census reveals a faster pace of growth in the dwelling counts than in the previous five years. The Australian Capital Territory, Victoria and Queensland experienced the highest rates of growth in the number of dwellings, and South Australia the lowest, as shown in Table 3.6.1. Of interest is the acceleration in dwelling growth seen in some of the states and territories such as New South Wales and Tasmania compared to the previous five years. In contrast, Western Australia and the Northern Territory have experienced a slowing of dwelling growth compared to the previous five years after a slowdown in the mining boom.

The occupancy rates for private dwellings are consistently higher in the 2021 Census than in the two previous censuses. Occupancy rates increased across all states, ranging from 88.7% in Tasmania to 93.6% in the Australian Capital Territory (see Table 3.6.2).

Changes in private dwelling occupancy are affected by several factors including the amount of the available dwelling stock; the number of people travelling and away from home on Census night; the number of overseas visitors and usual residents returning from overseas; and other demographic changes such as migration. Many of these factors have been impacted by the COVID-19 pandemic which saw movements of people, within, into, and out of Australia restricted.

Occupancy rates can also vary according to how well the ABS has determined the occupancy status of a dwelling, and in 2021 the ABS made two key innovations to enhance this determination. First, ABS made it easier for people to report that their home was not occupied on Census night. Secondly, a new occupancy indicator based on administrative data was created. This indicator was used during data processing to better identify whether a dwelling was occupied or not in the event that there was insufficient information from the field to do this. These innovations were introduced to address the previous Panel’s observation on occupancy determination (see Appendix A), however their introduction proved particularly beneficial given the impact of COVID-19 lockdowns on field enumeration activities in a number of jurisdictions and the inability to have personal contact.

The increases in occupancy may be somewhat understated because the number of dwellings deemed to be occupied in the 2016 Census was over-estimated. However, estimates from the Post Enumeration Survey indicate that there was reduced over-imputation for private dwellings in 2021 compared to 2016, nationally and in most states or territories. In other words, improvements have been observed in the correct identification of occupancy in the Census. Therefore, the increase at the national level reflects real changes between 2016 and 2021 Censuses, although slightly understated.

Separate houses represented 70.1% of the dwelling stock in 2021 continuing a decline from 71% in 2016 and 73.7% in 2011 (see Table 3.6.3). Conversely, flats or apartments in a four or more storey block increased their share from 5.4% of Australia’s dwelling stock in 2016 to 7.5% in 2021, a sign of growing urban intensification. There has been a reduction in the proportion of dwellings which had their structure type listed as ‘not stated’ down to 0.3% in 2021 from 0.5% in 2016.

The ABS introduced new processes and procedures for the Dwelling structure data item in the 2021 Census in response to feedback from the 2016 Independent Panel. These included making more use of an increasingly mature Address Register to populate Dwelling structure; the use of Building Approvals data to source Dwelling structure information for some new dwellings; and the development of an imputation model to assign Dwelling structure where no other data is available. This new imputation model was particularly important in 2021 with the introduction of the new self-service option of the ‘No-Census Number’ option (see Appendix E). This option allowed users to complete their Census without the code and password delivered to their address (a measure introduced to increase response), and resulted in an increase in the number of dwellings without a dwelling structure. Dwelling structure imputation addressed this issue where there was adequate geographic location information to use the imputation model (see Appendix A for more detail).

Flats or apartments attached to a house represented 0.2% of dwellings in 2021, an increase resulting from the ABS making use of administrative data to source information on these types of dwellings.

There has been very little change in housing tenure over the last three Censuses. Dwelling owned with a mortgage is the most common tenure type and this has remained steady, as shown in Figure 3.6.1. The proportion of dwellings owned outright declined slightly between 2011 and 2016 and has stayed at that level since. The proportion of dwellings being rented has remained steady since 2016 after a rise between 2011 and 2016.

There have been significant changes in the number of non-private dwellings by type (see Table 3.6.4). There are several reasons for these changes:

• Some non-private dwellings were temporarily closed on Census night because of the pandemic. This applied to hotels, motels, etc (-10.2%) and to a large extent to staff quarters (-16.0%). The decline in the number of persons in staff quarters was not as great suggesting it was the smaller quarters that were closed (e.g. staff quarters on agriculture establishments).
• Some non-private dwellings have closed or permanently changed. The decline in nurses’ quarters (16.1%) occurred on the frame prior to enumeration and reflects the closure of some establishments and a shift to their broader use for hospital staff and subsequent shift in classification.
• Some reflect improvements in the frame as a result of collaboration with state government and local government agencies. Boarding houses (+118.0%), hostels for the homeless and night shelters (+20.7%), and other welfare institutions (+44.2%) fall into this category.
• Some reflect real growth such as the 18.2% increase in residential colleges.
• The 39.0% decrease in hostels for the disabled is explained by the treatment of small hostels as private dwellings in the 2021 Census.

Overall, it appears that there was a more accurate frame for non-private dwellings in 2021 than in 2016.

The count of Census private dwellings aligns well with new housing supply estimates by ABS’s Construction Statistics. The increase in occupancy rates is what you would expect during a pandemic, especially with many parts of Australia in lockdown at the time of the Census. The changes implemented by the ABS appear to have improved the accuracy of this measure. The changes in dwelling structure type are in line with expectations and the reduction in the ‘not stated’ category is a positive outcome although it is higher than that of 2011 Census. The work undertaken to improve the non-private dwelling frame should lead to improved quality for that part of the Census.