Chapter 12 - Outputs and dissemination

Latest release
Wage Price Index: Concepts, Sources and Methods
Reference period


12.1 This chapter describes the WPI data published by the ABS. It also explains how to interpret index numbers, the differences between index points and percentage changes, and how to construct index series from the data. It concludes by discussing the reliability of the WPI outputs.

Published statistics

12.2 The WPI is compiled quarterly by the ABS. The survey reference date is the last pay period ending on or before the third Friday of the middle month of the March, June, September and December quarters each year. The data are typically released approximately seven weeks after the end of each quarter, in the publication Wage Price Index, Australia (cat. no. 6345.0).

12.3 The main mechanism for dissemination of ABS data is through the ABS website The website provides free of charge: 

  • the main findings from the statistical releases
  • a version of the publication in PDF format which may be downloaded
  • a range of time series spreadsheets containing all available indexes in Microsoft Excel format.


12.4 Four sets of wage price indexes are released:

  • ordinary time hourly rates of pay excluding bonuses
  • ordinary time hourly rates of pay including bonuses
  • total hourly rates of pay excluding bonuses
  • total hourly rates of pay including bonuses.

12.5 The "headline measure" of the wage price index is the index for total hourly rates of pay excluding bonuses. Separate indexes are released for each of the above series for various combinations of state/territory, sector (private/public) and industry divisions. Estimates are published quarterly and more detailed data are available on request. Seasonally adjusted and trend data are produced for the index of total hourly rates of pay excluding bonuses for Australia, for the private and for the public sector.

Quarterly and annual data

12.6 The WPI is published on a quarterly and financial year basis. The index number for a financial year is the simple arithmetic average (mean) of the index numbers for the four quarters of that year. Index numbers for calendar years are not published by the ABS, but can be calculated as the simple arithmetic average of the quarterly index numbers for the year concerned.

Release of WPI data

12.7 It is standard ABS policy and practice to make all our statistical releases available on our website to users of our statistics, simultaneously from 11.30 am (Canberra time) on the day of their release. Prior to 11.30 am, all ABS statistics are treated as confidential and regarded as 'under embargo'.

12.8 However, given the relatively high level of interest in the WPI, it is important from a 'public good' perspective that key ministers are able to respond in an informed manner to requests from the media for early comment on the released statistics. For this purpose, a secure 'lockup' facility is provided to enable authorised government officials and ministerial staff time to analyse the released statistics and develop a briefing to be provided to relevant ministers after lifting of the embargo.

12.9 Authorised persons attending a lockup are required to remain in a secure room managed by ABS staff, and are prohibited from communicating any information from the statistical release to anyone outside the room, until the embargo is lifted. Attendees at the lockup are also required to sign security undertakings which include provision for prosecution under the Crimes Act 1914 for anyone who breaches the conditions for attending the lockup. A list of products approved for provision to authorised persons via ABS-hosted lockups is available on the ABS website on the 'Policy on Pre-Embargo Access to ABS Statistical Releases' in the 'About Us' section.

Seasonally adjusted indexes

12.10 The WPI produces an original, seasonally adjusted and trend series for the index of total hourly rates of pay excluding bonuses for Australia. Adjusted series are available for the All, Private and Public sectors.

12.11 Seasonally adjusted estimates are derived by estimating and removing systematic calendar related effects from the original series. In most economic data these calendar related effects are a combination of seasonal influences (e.g. the effect of the weather, social traditions or administrative practices) and other kinds of calendar related variation, such as the number of trading days, Easter or the proximity of significant days in the year (e.g. Christmas). In the seasonal adjustment process, both seasonal and other calendar related factors evolve over time to reflect changes in activity patterns.

12.12 The total hourly rates of pay excluding bonuses index is the only index of the WPI that is seasonally adjusted. Institutional effects largely drive the seasonality of this index. Important factors in determining this seasonality are the timing of effect of agreements, the length of these agreements, and the timing of the implementation of significant wage determinations that impact on rates of pay. A significant institutional change in wage setting arrangements can affect the relative level (or trend) and seasonality of the index.

12.13 The WPI uses a concurrent seasonal adjustment methodology to derive the adjustment factors. This method uses the original time series available at each reference period to estimate seasonal factors for the current and previous quarters. Concurrent seasonal adjustment is technically superior to the more traditional method of reanalysing seasonal patterns once each year because it uses all available data to fine tune the estimates of the seasonal component each quarter. With concurrent analysis, the seasonally adjusted series are subject to revision each quarter as the estimates of the seasonal factors are improved. In most instances, the only significant revisions will be to the combined adjustment factors for the previous quarter and for the same quarter in the preceding year as the reference quarter (i.e. if the latest quarter is Qt then the most significant revisions will be to Qt-1 and Qt-4). Seasonal patterns are also reanalysed annually to reflect known changes to regular events. This can lead to additional revisions.

ARIMA modelling

12.14 The ABS uses Auto-Regressive Integrated Moving Average (ARIMA) modelling techniques to produce seasonally adjusted estimates. ARIMA modelling is a technique that can be used to extend original estimates beyond the end of a time series. The extended values are temporary, intermediate values that are used internally to improve seasonal adjustment. They do not affect the original estimates and are discarded at the end of the seasonal adjustment process. The use of ARIMA modelling generally results in a reduction in revisions to the seasonally adjusted estimates when subsequent data becomes available. ARIMA modelling in the WPI was introduced in the June quarter 2008. For more information on the details of ARIMA modelling see the feature article 'Use of ARIMA modelling to reduce revisions' in the October 2004 issue of Australian Economic Indicators (cat. no. 1350.0).

Trend estimates

12.15 Trend is a measure of the underlying direction of a series. The ABS trend estimates for the WPI are derived by applying a 7-term Henderson-weighted moving average to all quarters of the respective seasonally adjusted indexes except the first three and last three quarters. Trend estimates are created for these quarters by applying surrogates of the 7-term Henderson weighted moving average to the seasonally adjusted indexes, tailored to each time series. In general, trend estimates give a better indication of underlying behaviour than the seasonally adjusted estimates. Please refer to the ABS information paper, A Guide to Interpreting Time Series - Monitoring Trends (cat. no. 1349.0).


12.16 Original index numbers will be released as final figures at the time they are first published. Revisions will only occur in exceptional circumstances. Trend and seasonally adjusted indexes for some quarters will be updated as extra quarters are included in the series analysed for seasonal influences.

Interpreting index numbers

Index points and percentage changes

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

Total hourly rates of pay excluding bonuses, All Sectors, Australia, Index numbers, original:
June quarter 2012 = 112.2 
Less index number for June quarter 2011 = 108.2 
Change in index points = 4.0 
Percentage change = 4.0/108.2 x 100 = 3.7%

12.18 For most applications, movements in price indexes are best calculated and presented as percentage changes. Percentage change allows comparisons in movements that are independent of the level of the index. For example, a change of 2.0 index points when the index number is 120.0 is equivalent to a change of 1.7%. But if the index number were 80.0, a change of 2.0 index points would be equivalent to a change of 2.5%, a significantly different rate of price change. Only when evaluating change from the index reference period of the index will the points change be numerically identical to the percentage change.

12.19 The percentage change between any two periods must be calculated, as in the example above, by direct reference to the index numbers for the two periods. Adding the individual quarterly percentage changes will not result in the correct measure of longer term percentage change. That is, the percentage change between (say) the June quarter of one year and the June quarter of the following year will not necessarily equal the sum of the four quarterly percentage changes. The difference becomes more noticeable the longer the period covered, and the greater the rate of change in the index. This can readily be verified by starting with an index of 100.0 and increasing it by 10% (multiplying by 1.1) each period. After four periods, the index will equal 146.4 delivering an annual percentage change of 46.4%, not the 40.0% obtained by adding the four quarterly changes of 10.0%.

12.20 Although the WPI is compiled and published as a series of quarterly index numbers, its use is not restricted to the measurement of price change between quarters. A quarterly index number can be interpreted as representing the average price during the quarter (relative to the index reference base), and index numbers for periods spanning more than one quarter can be calculated as the simple (arithmetic) average of the quarterly indexes. For example, an index number for the calendar year 2011 is the arithmetic average of the index numbers for the March, June, September and December quarters of 2011.

Precision and rounding

12.21 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 for the total level of a group of indexes being outside the range of the percentage changes for the component level indexes. Seasonally adjusted and trend quarterly estimates are calculated from unrounded original indexes. The percentage changes (rounded to one decimal place) are calculated from the rounded index numbers.

Reliability of the indexes

Sampling error

12.22 Since the index numbers of the WPI are based on information relating to a sample of employee jobs, they are subject to sampling error. That is, they may differ from figures that would have resulted had all relevant employee jobs in the labour market been included in the collection. While it is reasonably straightforward to calculate sampling errors for a level estimate such as the total number of employee jobs estimated via a sample survey, it is not so straightforward a process for the WPI which is a product of sampling and index methodologies. The ABS has not published any estimates of sampling error for the WPI.

Non-sampling error

12.23 Inaccuracies in the data may also occur because of imperfections in reporting by businesses or in processing by the ABS. Inaccuracies of this kind are referred to as non-sampling errors. Every effort has been made to reduce non-sampling error, for example:

  • by careful design and testing of questionnaires and processing systems
  • by providing instructions to businesses on how to select a sample of employee jobs
  • by detailed checking of completed survey forms
  • by instituting a range of procedures for ensuring that jobs are priced to constant quality.
Back to top of the page