Australian Bureau of Statistics

Rate the ABS website
ABS Home > Statistics > By Catalogue Number
ABS @ Facebook ABS @ Twitter ABS RSS ABS Email notification service
6464.0 - House Price Indexes: Concepts, Sources and Methods, Australia, 2009  
Latest ISSUE Released at 11:30 AM (CANBERRA TIME) 14/12/2009   
   Page tools: Print Print Page Print all pages in this productPrint All RSS Feed RSS Bookmark and Share Search this Product

CHAPTER 8 SAMPLING


INTRODUCTION

8.1 The HPI measures price change of the stock of established houses in the eight capital cities. Ideally, the HPI would be compiled using the current and historical market prices of the entire stock of houses. In practice, market prices for any particular period are only available for those houses that are actually traded (sold/purchased) in the period. Such sales account for only a very small proportion of the total housing stock in any quarter and so it is necessary to draw inferences about the price behaviour of the whole stock from these small samples. The assumption behind this procedure is that the median sales price of the houses traded each quarter is indicative of the median price of all houses.

8.2 To allow the compilation and publication of a more timely estimate of price change, the preliminary estimates of the two most recent quarters in the series contain a sample of mortgage lenders' data to supplement the available data from Valuers-General (VGs), which is provided progressively over time as properties are settled and transactions are registered. The index is therefore not a measure of every house sale transaction in a given period. This sampling approach is consistent with methodological practice across the suite of ABS price indexes.

8.3 The preliminary estimates for the two most recent quarters, referred to as the leading indicator series, are revised as more data are progressively received from the VGs. The third most recent quarter in any publication is calculated from VGs data only and will not be subject to further revision once published. The index numbers derived only from VGs data are referred to as the benchmark series.


PROVIDERS

Valuers-General

8.4 Each property transaction, regardless of type or location, is registered to enable the relevant State government authority to maintain a record of property ownership and to facilitate other legislated functions, such as land valuations for council ratings, and the collection of taxes and duties.

8.5 The most obvious source of comprehensive information on house prices are the State Valuers-General offices. The data held by these agencies represent the ABS’ preferred source for compiling the HPI because they provide the most comprehensive dataset currently available on house sales. The information contained in these records varies between jurisdictions. While all include details of the transaction (date, price, etc.), in some jurisdictions, information about the physical characteristics of the property is also available.


Mortgage Lenders

8.6 A more timely data source which is used to supplement the VGs data is property loan applications from mortgage lenders. A large percentage of house sales involve mortgages, and such applications are generally processed shortly after the exchange of contracts. Loan documents and the systems used by most mortgage lenders do not capture the actual date of exchange. However, the recorded loan application approval date has proven to be a satisfactory proxy for the date of exchange.

8.7 While not all mortgage lenders are included in the HPI sample, the coverage does include major banks and is representative of the market.

8.8 All records relating to individual transactions are treated with confidentiality and stored securely. Only aggregate information (e.g. price indexes) is released.

8.9 It should be noted that VGs and mortgage lenders are included in the HPI sample on an on-going basis. That is, there is no 'rotation' of providers, which reduces the complexities that otherwise would be involved with managing the compositional change that would arise.


LIMITATIONS

8.10 A disadvantage of the VGs administrative datasets is the lengthy delays experienced before all data become available. Different jurisdictions have different legislation governing the reporting requirements of parties to property transfers. In general, the requirement is for the property transfer to be registered within 60-90 days of settlement. When combined with a lag between exchange of contracts and final settlement of 4 to 6 weeks on average, but up to 3 or 4 months in some cases, the delay between the date of contracts being exchanged and all transactions relating to a particular month being received by the ABS can be 6 months or more.

8.11 A further concern about obtaining reliable price measures is that properties with higher prices generally take longer to settle. The consequence is that details received by the ABS relating to the property sales in a particular quarter are distributed in a biased way. In general, the median price of properties exchanged in a particular quarter increases as the dataset becomes more complete. The resulting bias in early reported data is always downwards but its magnitude is not consistent, either between cities or over time within any one city. As a result, it is necessary to obtain an almost complete dataset for each quarter before it is possible to determine the most accurate measure of median house prices for each cluster. As noted above, it takes several months for all transactions relating to a particular quarter to be finally settled, recorded by the relevant State agency and then passed on to the ABS.

8.12 Some mortgage lenders’ data also have a shortcoming in that loan documents do not necessarily record the actual sale price of the property, rather these records sometimes contain the security valuation amount, which can differ markedly from the sale price. Though the most obvious of these records are identified and excluded, the median prices derived from mortgage lenders data can differ from median prices derived from the complete VGs dataset.

8.13 The additional data provided by the sample of mortgage lenders allow compilation and publication of a more timely estimate of price change. The preliminary estimates for the two most recent quarters, the leading indicator series, are therefore revised as more VGs transaction data are progressively received. Where the same sales are recorded in VGs and mortgage lenders' data, the VG record is used.


BIASES

8.14 As stated above, different States have different legislation in place concerning the length of time in which an owner must register the property title transfer. Further, there is a bias in early VGs data caused by the tendency of properties with higher prices to take longer to settle and therefore appear in the VGs dataset. The VGs data available for the two most recent quarters are biased downwards because of this tendency for cheaper properties to be settled more quickly than relatively expensive properties.


Reducing biases

8.15 The method used to calculate price relatives serves to counter the effects of this bias on the composition of the sets of price data collected every quarter. Price relatives are determined only by comparing current benchmark medians (BM) with previous benchmark medians and current leading indicator medians with previous leading indicator medians. Thus, in the leading indicator series, medians from the current second preliminary estimates (P2) quarter are compared with medians from the previous second preliminary estimates quarter, and medians from the current first preliminary estimates (P1) quarter are compared with medians from the previous first preliminary estimates quarter. Using this approach, price relatives are derived from datasets which have a similar proportion of VGs and mortgage lenders' data.

8.16 In a particular processing cycle where period t is the most recent quarter, the price relatives for the benchmark and the two leading indicator quarters can be represented algebraically as follows:

Equation: Price relative of benchmark = P(i, t-2, BM)/P(i, t-3, BM); Price relative of P2 = P(i, t-1, P2)/P(i, t-2, P2); Price relative of P1 = P(i, t, P1)/P(i, t-1, P1);

8.17 For example, suppose that September quarter is the most recent quarter in an index production cycle, period t. September quarter will be the first preliminary estimate (P1) of the indicator series, June quarter (period t-1) will be the second preliminary estimate (P2) of the indicator series, and March quarter (period t-2) will be the benchmark series (BM). In the previous production cycle June quarter was P1, March quarter was P2 and December quarter (period t-3) was benchmark. The September quarter price relatives will be calculated by dividing the September quarter P1 median by the June quarter P1 median. The June quarter price relatives will be calculated by dividing the June quarter P2 median by the March quarter P2 median. The March quarter price relatives will be calculated by dividing the March quarter BM median by the December quarter BM median.


Bookmark and Share. Opens in a new window

Commonwealth of Australia 2014

Unless otherwise noted, content on this website is licensed under a Creative Commons Attribution 2.5 Australia Licence together with any terms, conditions and exclusions as set out in the website Copyright notice. For permission to do anything beyond the scope of this licence and copyright terms contact us.