Residential Property Price Indexes: Eight Capital Cities methodology

This release contains a range of residential property price indexes and related statistics.

Residential property prices are of significant interest to policy makers, market analysts and researchers for a range of economic and social reasons. This is because the housing market plays an important role in the Australian economy.

Residential property price indexes measure price change of the stock of residential dwellings over time.

The indexes serve the dual purpose of:

• a macroeconomic indicator of residential property price inflation
• supporting the compilation of the non-financial assets component of the household balance sheet in the Australian System of National Accounts.
   

The suite of residential property price indexes (from now on referred to collectively as 'the indexes') is:

• a residential property price index
• an established house price index
• an attached dwellings price index.
   

The residential property price index is an aggregation of the established house price index and the attached dwelling price index, and measures the price change in all residential dwellings within the eight Greater Capital City Statistical Areas (from now on referred to collectively as 'the capital cities'). Index numbers and percentage changes for the residential property price index are presented in Table 1.

The established house price index measures the price change in all established detached houses on their own block of land in the capital cities. Index numbers and percentage changes for the established house price index are presented in Table 2.

The attached dwellings price index measures the price change of attached dwellings in the capital cities. Dwellings in scope of the index are:

• flats, units and apartments
• semi-detached, row and terrace houses.
   

Index numbers and percentage changes for the attached dwellings price index are presented in Table 3.

Estimates are also available of median price and transfer counts of established houses and attached dwellings, for capital cities and rest of state (Tables 4 and 5). A transfer is defined as a change of ownership.

Estimates of the total value of all residential dwellings are presented in Table 6. Values of dwellings and land are used in compiling the non-financial assets component of the household balance sheet published annually in the Australian System of National Accounts and quarterly in Australian National Accounts: Finance and Wealth.

The scope of the indexes:

• all residential properties in the eight capital cities
• restricted to dwellings where the primary purpose is residential, regardless of ownership and tenure of the occupants
• includes government owned properties and properties owned by private landlords
• excludes commercial properties
   

Dwellings in scope of the indexes are:

• Ordinary detached houses
• Houses with offices
• Houses with flats
• Rural residential houses (within a capital city and not part of a farming business)
• Semi-detached, row and terrace houses
• Townhouses
• Flats, units and apartments.
   

The price indexes and related statistics are compiled from analysis of a quarterly dataset of all Australian residential property sales. The sale price is the price agreed when contracts for sale are exchanged. The exchange date most closely approximates the time at which the market price is determined.

Sales are attributed to the relevant quarter depending on the date of exchange of contracts. Exchange date information is available for all cities except Adelaide and Darwin. For these cities, a modelled exchange date is used.

The definitions of dwelling structure type are consistent with the relevant ABS classifications:

• the Functional Classification of Buildings 1999 (Revision 2011), which is used in building activity statistics
• the Dwelling Structure Classification, which is used in the Census of Population and Housing (refer to Census Dictionary, 2016).
   

Capital cities are defined as Greater Capital City Statistical Areas, as described in the Australian Statistical Geography Standard, Vol 1. These areas capture the socio-economic extent of the state/territory capital cities for statistical purposes. This definition applies from the December quarter 2013 issue of this publication.

For earlier periods, capital city regions are defined in terms of Statistical Divisions under the Australian Statistical Geography Classification. Historical naming conventions (for example, Sydney rather than Greater Sydney) have been maintained. A time-series is available but users should exercise caution in interpreting medians and numbers of house transfers over time, as historical data will reflect capital city boundaries as previously defined. This is particularly significant for Canberra, where the capital city is now defined to be the whole of the ACT.

The portion of each state or territory outside its capital city region is referred to as the rest of state.

Where table headings indicate that the estimates relate to the rest of state or whole of state, the Australian Statistical Geography Standard is used to determine boundaries. For example, the total value of the dwelling stock relates to each state or territory.

All Australian residential property sales data are supplied to the ABS by CoreLogic. (See Appendix: CoreLogic disclaimer and copyright notices). This dataset is a combination of residential property sales data obtained from state and territory land titles offices or Valuers General offices, and real estate agents' data provided to CoreLogic. The ABS applies classifications to this dataset to create the residential property sales dataset from which the price indexes and related statistics are produced.

The ABS employs a stratification approach to compile the indexes. Stratification involves separating the dwelling stock population into non-overlapping groups, called strata, which are defined by dwelling type, long term median price and Socio-economic Index for Areas score. The sample of dwelling transactions within each stratum is used to estimate the stratum's quarterly price change.

Each quarter, the strata are re-valued using a price relative. For each stratum, the price relative is the current median price compared to the previous median price. The price relative is applied to the value of the dwelling stock for that stratum in the previous period, to produce a current period stratum value. The current period values of each stratum are then summed to derive the current value of the total dwelling stock in the capital city. Index numbers are subsequently derived from the total values.

When the number of price observations available for a stratum is nil or extremely low in a quarter, a price movement for the stratum is derived using imputation methods based on price movements of other strata. See below for further details regarding imputation.

Information on the stratification methodology is available in Residential Property Price Indexes: Concepts, Sources and Methods.

The weights underpinning the indexes are based on the total value of dwellings (including land) in scope of the indexes. The weights are updated at roughly five yearly intervals to take account of changes in the quantity (number) of dwellings. Dwelling counts are obtained from the five yearly Census of Population and Housing, and are combined with mean-adjusted median prices calculated from Valuers General data to produce new weights for the indexes. The most recent weights are published in the December quarter 2018 issue.

Stratification methods require a sufficient number of residential property transactions to compile the indexes each quarter. A substantial fall in the numbers of residential property transactions (for example, due to the effects of COVID-19) can render the transaction prices which are observed less representative of the stock of dwellings in a city. To overcome this, the ABS uses imputation methods to estimate price changes in some strata and cities with low transaction counts.

The range of available imputation methods is outlined below:

• Donor imputation is used where the number of residential property transactions in the quarter is extremely low for the sample to provide a reliable estimate of price change, and a suitable ‘donor’ series is available for imputation. This approach imputes a price movement from a donor series such as another stratum, dwelling type or city.
• Mean imputation is used where the number of residential property transactions in the quarter is nil or extremely low for a stratum to provide a reliable estimate of price change, and a suitable donor series is not available for imputation. This approach derives the stratum movement using the average (mean) price movement of all other strata for that dwelling type in that city.
• Carry forward imputation imputes a zero price movement, by carrying forward the previous quarter's median price.
• Headline imputation is used where the number of residential property transactions in the quarter is extremely low for a city to provide a reliable estimate of price change. This approach imputes a price movement for a city index using the movement of the 'headline' index calculated on the remaining cities. This has the effect of the movement of the imputed series not contributing to the headline quarterly movement.
• Unstratified median imputation imputes a price movement for a city index using the movement in that city's median price calculated without stratification.
   

Each of these options may be suitable in different circumstances. The underlying principle is that when imputation is required, the imputed price movement is expected to provide a reliable estimate of price change for the series being imputed.

The number of price observations available to compile the indexes each quarter depends on market activity. For the smaller capital cities (Hobart, Darwin and Canberra) strata sometimes have low numbers of price observations. Rather than suppress publication of the series, they are included, as the long term trends are considered reliable. Care should be exercised when analysing the indexes' quarter-to-quarter movements for the smaller capital cities.

The published index numbers have been rounded to one decimal place, and the percentage changes (also rounded to one decimal place) are calculated from the rounded index numbers. In some cases, this can result in the percentage change in an aggregate index (the residential property price index) being outside the range of the percentage changes in its component level indexes (the established house price index and the attached dwellings price index).

Indexes are released as final figures when first published. Revisions would occur only in exceptional circumstances, such as to correct a significant error.

Estimates of the total value of the dwelling stock are available in Table 6 and comprise three outputs:

• the mean price of residential dwellings
• the number (or quantity) of residential dwellings
• the total value of residential dwellings (which is an aggregation of the price and quantity components).
   

Dwellings in scope of the value of the dwelling stock are the same as for the indexes; however, geographic coverage is expanded to the whole of state. From the dataset provided by CoreLogic, only the residential property sales data obtained from state and territory land titles offices or Valuers General offices are used in estimating the total value of the dwelling stock. Real estate agents' data provided to CoreLogic are excluded.

As with the price indexes, the total value of the dwelling stock uses a stratification approach. Price, quantity and value information is stratified by location (in terms of Statistical Area Level 2 in the Australian Statistical Geography Standard) and dwelling type (established houses and attached dwellings).

A representative price for all dwellings in the stock is obtained from information on dwellings sold during the reference period. Price information from dwellings sold is used to represent the price of all dwellings not sold during the period. A quarterly mean dwelling price by geographic area and by dwelling type for all strata is calculated.

The number of residential dwellings is calculated by taking counts of dwellings from the latest Census and adjusting these counts for net additions to the stock since the last Census. These net additions are calculated by taking completions data from Building Activity, Australia and adjusting the completions data by the long term realisation rate (the long term average rate at which completions result in net additions to the stock).

The total number of residential dwellings is calculated at the state level and pro-rated down to each Statistical Area Level 2. As completions data are not available in time for use in compiling the most recent quarter's estimates, and are not available below state/territory level, quantity information for the latest quarter is modelled using historical trends.

To compile the total value of dwelling stock, price and quantity data are combined in each Statistical Area Level 2 and then aggregated up to the state/territory and national level. Information from the Census is used to further break down total value information into Household and Non-Household sector ownership.

For the total value of the dwelling stock, mean prices are calculated from unrounded figures and subsequently rounded. Therefore, estimates of the components published in Table 6 cannot be combined to replicate the total values.

To enable the timely publication of data on the value of the dwelling stock, the movement of the residential property price index (at the capital city level) is used as a proxy for movements in the mean prices (at the state/territory level) for the most recent quarter. This results in the estimates being preliminary in this period.

Revisions to the total value of the dwelling stock are applied to both the second and third most recent quarters, due to updated completions and unit record sales data becoming available to use in the direct calculation of quantity and price. The estimate for the third most recent quarter is considered final.

Once the estimates are final, revisions would occur only in exceptional circumstances, such as to correct a significant error.

Further information on the methodology used to compile the total value of the dwelling stock is available in the Residential Property Price Indexes: Concepts, Sources and Methods.

In addition to the stratified and weighted price indexes for each capital city, the ABS publishes in Tables 4 and 5, for each capital city and the rest of state:

• the median price of all established house and attached dwelling transfers
• the number of such transfers.
   

The number of transfers of established houses and attached dwellings provides an indication of the level of sales activity for each quarter.

The median prices are calculated with no stratification or weighting applied. These 'raw' medians will not correspond to the published index numbers, and will not produce price movements that are consistent with those numbers.

As the ABS receives more unit record residential property sales data for each quarter, the median prices and the number of house and attached dwelling transfers are revised as necessary. Revisions can be made to the most recent ten quarters of published figures.

These revisions can be substantial, especially in the quarter immediately following first release, and accordingly the median price and transfer data initially published for each quarter should be interpreted with caution. For this reason, the current quarter median price and number of transfers are annotated as being preliminary and that the series is subject to revision.

The data are released about twelve weeks after the end of the quarter to which they relate.

From the March quarter 2015 issue the source of unit record residential property sales data has changed. The complete CoreLogic unit record file, including real estate agents' data, has been used to calculate the series from the March quarter 2014 up to the current quarter. There are differences in the numbers produced using this data source compared to the previous data source (which contained Valuers General data only). Therefore, users should apply caution when comparing data prior to March quarter 2014 with data for the March quarter 2014 and later periods.

An updated residential property sales dataset supplied by CoreLogic was implemented from the December quarter 2018 issue. This has resulted in revisions to the median price and transfers series. The ABS has revised these series back to March quarter 2014, when CoreLogic data was introduced into the series.

Please also see the above discussion of revisions, for each group of outputs, under How the data is processed. For a discussion of comparability of data over earlier periods than March 2014, please see Residential Property Price Indexes: Concepts, Sources and Methods.

Users should exercise caution in comparing price movements in the indexes and changes in the value of the dwelling stock and its components. The indexes are designed to measure the change in value of the stock of dwellings in the capital cities fixed at the last Census, whereas the total value of dwelling stock is designed to measure the current value of the dwelling stock in the states and territories. As such, movements in the value of the dwelling stock are a result of changes in both the price and quantity of dwellings. Movements in the indexes represent price change only.

Users should exercise caution when comparing the unstratified median prices published in Table 4 and the mean value of dwellings published in Table 6. The unstratified median price (for established houses and attached dwellings) of dwelling transfers over the reference period is the mid-point of all properties sold in the period.

This means that half of all properties (in the same region and of the same dwelling type) sold in the period did so at a price below the median, and the other half sold at a price above the median. In contrast, the mean value of residential dwellings represents the average dwelling value in the reference period.

The mean value is derived by taking the total value of residential dwellings and dividing by the estimated number of dwellings in the stock.

The mean values are calculated across the whole of state and for all dwelling types, whereas the medians are calculated for individual dwelling types and for the capital city and rest of state separately.

Movements in indexes from one period to another can be expressed either as changes in index points or as percentage changes. The following example illustrates the method of calculating index points changes and percentage changes between any two periods:

Established Houses: Sydney index numbers (see Table 2) -
June Quarter 2020 172.7
less March Quarter 2020 177.4
equals change in index points -4.7
Percentage change -4.7/177.4 x 100 = -2.6%

In this publication, percentage changes are calculated to illustrate two different kinds of movements in index numbers:

• movements between corresponding quarters of consecutive years
• movements between consecutive quarters.
   

Index numbers for financial years (not included in this publication) can be calculated as simple, arithmetic averages of the four quarterly index numbers for the financial year. The following example illustrates the method of calculating the financial year index number for 2019-20:

Established Houses: Sydney index numbers (see Table 2) -
September Quarter 2019 164.4
plus December Quarter 2019 172.9
plus March Quarter 2020 177.4
plus June Quarter 2020 172.7
equals 687.4
Financial year 2019-20 Index 687.4/4 =171.9

For more detailed information on residential property price indexes and related statistics, refer to Residential Property Price Indexes: Concepts, Sources and Methods.

Current publications and other products released by the ABS are listed on the ABS website https://www.abs.gov.au. The ABS also issues a daily Release Advice on the website which details products to be released in the week ahead.

The ABS has compiled a house price index since 1986. A significant review of the index occurred in 2004. Several improvements were implemented as a result of this review and a new series (Series 1) of the index was introduced in the September quarter 2005 issue (with improvements applied back to the March quarter 2002). The most significant change was the introduction of a stratification approach to compile the house price index. (See above, under How the data is processed, for a detailed description of the index methodology including the stratification approach.) For more information on the 2004 review, see Information Paper: Renovating the Established House Price Index.

The historical series, from 1986 to 2005, continues to be available as an indicator of established house price movements over a longer period. This historical series is not directly comparable to the existing house price index series post 2002, due to the change in methodology resulting from the 2004 review.

A second review of the house price index was undertaken in 2007. This review refined the stratification method and updated the dwelling stock values using 2006 Census data. Outcomes of the 2007 review were introduced in the December quarter 2008 issue, creating Series 2, which was linked to Series 1 at the March quarter 2008.

The next review, in 2012, resulted in the expansion in scope beyond the existing house price index to include attached dwellings and produce an aggregate residential property price index. The dwelling stock values were also updated using data from the 2011 Census. This third series (i.e. Series 3) was introduced in the December quarter 2013 issue and linked to Series 2 at the March quarter 2013. The index reference period for all indexes was also updated to 2011-12 = 100 in the December quarter 2013 issue.

In 2014, the residential property price index work program was reviewed in response to planned reductions to the ABS work program. The outcomes of this review were implemented in the March quarter 2015.

In summary, the changes from the 2014 review were:

• all Australian residential property sales data used to compile the price indexes and related statistics would now be supplied to the ABS by CoreLogic
• from the March quarter 2015 the suite of residential property price indexes are considered final when first released
• the method of calculating prices in the total value of the dwelling stock was modified, due to the change in timing of the release
• the unstratified median price and number of dwelling transfers series would now be published up to the current quarter.
   

This review did not result in the introduction of a new index series. For more information on the implementation of the outcomes of the review please refer to the Feature Article: Implementation of the ABS Residential Property Price Index Review Outcomes in the March quarter 2015 issue.

The most recent review of the residential property price index and the total value of the dwelling stock was conducted in 2018, with updates introduced in the December quarter 2018 issue. Outcomes of this review include:

• the weights and variables used to stratify the indexes were updated using data from the 2016 Census
• inputs used to compile the total value of the dwelling stock were updated using data from the 2016 Census
• changes to the residential property sales dataset supplied by CoreLogic were implemented.

The new price index series (Series 4) was linked to the existing series (Series 3) at the September quarter 2018. Updates to the total value of the dwelling stock were incorporated into the series over a number of quarters. A two year window was used to smooth in the change for the household series, while a four year window was used for the non-household series. Revisions to the unstratified median price and number of dwelling transfers series resulting from the changes to the residential property sales dataset have been incorporated from March quarter 2014, when CoreLogic data was introduced into this series.

Details of the concepts, sources and methods underlying the compilation of the current index series (Series 4) are described in the Residential Property Price Indexes: Concepts, Sources and Methods.