They are compositional changes, quality improvements and timeliness of the data available (RBA, 2004). These issues are not unique to Australia as similar challenges can be found in most advanced economies in the world as well. We take each one in turn below.
2.1 Compositional changes
At any given point in time, only a certain segment of the property market is transacted. Thus, the type of properties being transacted at any given time could drive price changes if compositional changes between periods are not considered. For example, if there are more 4-bedroom houses sold in the current period but there were more 1-bedroom units sold in the period before, the average or median price could see a significant increase even if no other price drivers have changed.
2.2 Quality improvements
Over time, properties in the national dwelling stock are constantly being improved to meet housing preferences. Improvements could include upgrades in the form of additional bedrooms, bathrooms or additional features that add to the value of the property itself. Thus, these improvements will be captured over time through a property price measure.
2.3 Timeliness
Unlike financial market data, property market transactions data are often only available at the point of settlement rather than contract signing. The Reserve Bank of Australia states that this is likely due to the decentralised nature of the market and the data relied on is often government administrative data, which is available only after settlement (RBA, 2004). Therefore, the most recent data will capture property market activity with a delay. This is also known as the settlement lag.
The settlement lag is directly related to the length of time between the contract signing and the point at which monies are transferred and keys are handed over, i.e. the settlement period. The length of a settlement period can be determined by a variety of factors which are unique to buyers’ and sellers’ circumstances. According to Asabere & Huffman (1993), such factors can include potentially having a longer than normal search for financing and delays on the sale of an existing house. They note that a longer settlement period poses an opportunity cost for the seller but may be beneficial if the seller is an owner-occupier who expects that it will take a while for them to find their next property to live in. The length of the settlement will likely depend on the complexity of the transaction to allow for the completion of all legal and financial paperwork and due diligence prior to settlement, as there are often severe financial penalties for not complying within the stated timeframe.
According to Canstar (2022), the average settlement period is between 30 and 90 days in Victoria and most other states and territories across Australia. This is consistent with the findings of a joint research project conducted by PEXA and Domain which found that the lag between buyer demand (defined by database searches and settlement activity) was approximately eight weeks in NSW and 11 weeks in Victoria (PEXA Insights, 2021).1 But while a typical transaction can settle anywhere between 30 and 90 days, the settlement period can have a long tail as there is a segment of these transactions that is highly complex and will require more time for settlement to occur. Additionally, off‑the-plan sales will also have a longer settlement period relative to sales of existing dwellings. The settlement period for off-the-plan apartments can be up to two to three years according to Shoory (2016).
Given that policy and financial decisions are based on the information set available at a particular point in time, revisions to the information set that significantly change inferences or alter desired policy outcomes could be problematic. To our knowledge, other than PropTrack (2022) there have not been any studies that have attempted to determine whether observing only partial snapshots in early data releases (or vintages) significantly change inferences based on final vintage data (i.e. when all transactions have been received). To conduct such a study, analysis will need to rely on a framework that allows for both time and the information content to be studied.
A burgeoning area in literature focuses specifically on real‑time economic and fiscal data to understand the impact of data revisions on forecasting, monetary policy and fiscal policy. Croushore (2011) provides a review of this literature on the topic of data revisions on monetary policy, specifically. It is with this framework in mind that the analysis is undertaken in our report.
In our investigations of the data vintages, we segment the property market in three ways. Firstly, through a location lens which is in line with the Alonso-Muth-Mills (Alonso, 1964; Muth, 1969; Mills, 1967; Wheaton, 1974) model. This differentiates properties according to their distance to the central business district (CBD) of a city. Secondly, by differentiating types of property, between houses (detached) versus units (non-detached). And thirdly, by the contract price itself. We look at the median contract price for each vintage to determine whether higher or lower value property sales for a given contract period make up a larger share of early vintages.
There are equally probable reasons for why each of these segments may or may not have a shorter settlement period and may be captured in the earlier vintages. Therefore, we go into this investigation without any priors.
Footnotes
[1] This analysis is however only limited to transactions that were originally posted on the Domain website that settled through the PEXA system.
Updated