6150.0 - Australian Labour Account: Concepts, Sources and Methods, Oct 2019

ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 12/11/2019

Page tools: Print

Print Page Print all pages in this product

Australian Labour Account methods

Compilation methods

The Australian Labour Account data tables are compiled using different methods, namely interpolation, extrapolation backcasting and benchmarking. Methods chosen are based on two factors: the context in which the data were originally collected, and ability to fill data gaps between collection points or reference periods.

Interpolation

Interpolation is a method of constructing new data points within the range of a discrete set of known data points. Where interpolation is used in the Australian Labour Account, it is generally designed to create a quarterly series between two annual data points when data with a quarterly frequency are not available. An example of this is measuring the number of public sector jobs, where quarterly data are estimated from two annual data points.

Extrapolation

Extrapolation is the process of estimating values of a variable beyond its original observed range. Some estimates in the labour account are derived by extrapolating data using various indicators, as information necessary to compile a comprehensive and complete account may not be sufficiently timely. For example, as there is a time lag between the current reference period and the release of Government Finance Statistics (GFS), data for employment subsidies in the Australian Labour Account are extrapolated forwarded based on the movements of previously observed data.

Backcasting

Backcasting is the process of estimating values of a variable prior to its original observed range, usually through analysing the growth rates or proportional distribution of a conceptually related series. In addition, some estimates for earlier time periods in the Australian Labour Account are backcast from partially observed information. For example, data from the Job Vacancies Survey are not available on the current industry classification prior to 2009, however the total number of job vacancies is known. Data on the current industry classification for earlier time periods have been backcast using by applying a concordance between the previous and current industry classifications, with the additional constraint that the known total number of job vacancies must be maintained.

Benchmarking

Benchmarking is the processes of combining sub-annual (quarterly) indicator data and annual data, and aligning them to produce quarterly economic data that combine the robustness of the annual ‘benchmark’ source while reflecting the pattern of sub-annual movement. Benchmarks (or annual data) are usually of higher quality because they come from annual surveys, which draw on more complete source data (e.g. balanced and audited company financial accounts), conduct more detailed enquiries, and generally have larger sample sizes. To create a quarterly series, the annual data provides the overall levels, to which a conceptually related quarterly indicator series is benchmarked. An example of this in the Australian Labour Account is estimating private sector filled jobs by benchmarking quarterly jobs data to annual data.

There are a number of methods used to benchmark flow data, depending on the type of data to be benchmarked. The method used the majority of the time, due to its accuracy and ease of implementation, is the ‘Proportional Denton Method’. This method preserves the movement of the quarterly data by minimising the absolute difference in the relative adjustments of two neighbouring quarters (i.e. keeping the benchmarked data to indicator data ratio as constant as possible over the time series), under the constraint that the sum of the quarters is equal to the annual data for each benchmark year.

The Australian Labour Account uses a modified Proportional Denton Method to benchmark the Quarterly Business Indicators Survey (QBIS) industry data to the annual industry data from the Economic Activity Survey (EAS).

The standard Proportional Denton Method is modified for use in the Australian Labour Account in the following ways:

the Proportional Denton Method is generally used only in relation to flow data. In the Australian Labour Account, the mathematics underlying the Proportional Denton method have been modified to apply to stock data;
the Proportional Denton Method is generally not used to extrapolate data series beyond their observed range. In the Australian Labour Account, annual industry data from the EAS, which are not yet available, have been extrapolated for the latest year as part of the modified Proportional Denton Method by assuming a benchmark data to indicator data ratio of one;
in the context of flow data, the annual benchmark data measures a variable over an entire year and so should (theoretically) be equal to the sum of the four indicator data points for that year. In contrast, stock data measure a variable at a single point in time, and the annual stock benchmark data could simply be considered a more accurate measure of the indicator data of that quarter. The modified Proportional Denton Method used in the Australian Labour Account imposes an additional constraint for stock estimates, that the benchmarked quarterly data must be equal to the annual benchmark data in the June quarter of each year while maintaining, as much as possible, the quarterly movements of the indicator data.

For more information regarding the Proportional Denton Method, refer to paragraph 7.40 in the Australian System of National Accounts: Concepts, Sources and Methods (cat. no. 5216.0).

Annual Australian Labour Account data

Data in the Australian Labour Account are compiled with quarterly estimates as the primary level of data compilation, with annual estimates subsequently produced from quarterly data. The method used to annualise data varies for each quadrant, depending on whether data are stock or flow estimates.

Stock data

The Jobs and Persons quadrants contain stock data, which are data that measure certain attributes at a point in time. Data in these quadrants are annualised using a simple arithmetic average of the four quarterly estimates. While these average annual levels are not true stock values, in the sense that they are not measured at a specific point in time, the purpose of presenting annual estimates as an arithmetic average is to minimise issues with using any particular quarterly observation to represent an annual stock, as any particular quarterly observation may under or over represent “usual” stock levels for a particular year. This is particularly relevant for industries which exhibit strongly seasonal employment levels, for example retail trade.

For example, consider the example in Table 6.1 below of two industries which exhibit the following patterns in employed persons over a one year period.

Table 6.1 – Annual stocks example


Time period	Industry A – employed persons (000’s)	Industry B – employed persons (000’s)

Sep-15	115	220
Dec-15	120	300
Mar-15	125	230
Jun-15	130	220
2015-16 annual average	123	243

The annual average stock level for 2015-16 for Industry A is 123 thousand employed persons. The choice of using an annual average, an end of year stock level (of 130 thousand employed persons) or a mid-point stock level (of 120 thousand employed persons) for this industry does not significantly change the annual level of employed persons.

For Industry B, which shows a strong cyclical increase in employed persons each December, the choice of annual stock level is more significant. If an annual average stock level (of 243 thousand employed persons in 2015-16) or end of year stock level (of 220 thousand employed persons) were chosen, a much lower annual stock level would result than if a mid-point stock level (of 300 thousand employed persons) were used.

Flow data

The Labour Volume and Labour Payments quadrants contain flow data, which represent a measure of activity over a given period. Data in these quadrants are annualised as the sum of the four quarterly estimates.

Seasonal adjustment

Any original time series can be thought of as a combination of three broad and distinctly different types of behaviour, each representing the impact of certain types of real world events on the information being collected: systematic calendar related events, short-term irregular fluctuations and long-term cyclical behaviour.

Seasonal adjustment is a statistical technique that attempts to measure and remove the effects of systematic calendar related patterns including seasonal variation to reveal how a series changes from period to period. Seasonal adjustment does not aim to remove the irregular or non-seasonal influences, which may be present in any particular data series. This means that movements of the seasonally adjusted estimates may not be reliable indicators of trend behaviour.

The ABS software for seasonal adjustment is the SEASABS (SEASonal analysis, ABS standards) package, a knowledge based seasonal analysis and adjustment tool. The seasonal adjustment algorithm used by SEASABS is based on the X-11 Variant seasonal adjustment software from the U.S. Census Bureau.

Trend estimates

In cases where the removal of only the seasonal element from an original series (resulting in the seasonally adjusted series) may not be sufficient to allow identification of changes in its trend, a statistical technique is used to dampen the irregular element. This technique is known as smoothing, and the resulting smoothed series are known as trend series.

Smoothing, to derive trend estimates, is achieved by applying moving averages to seasonally adjusted series. A number of different types of moving averages may be used; for quarterly series a seven term Henderson moving average is generally applied by the ABS. The use of Henderson moving averages leads to smoother data series relative to series that have been seasonally adjusted only. The Henderson moving average is symmetric, but asymmetric forms of the average may be applied as the end of a time series is approached. The application of asymmetric weights is guided by an end weight parameter, which is based on the calculation of a noise-to-signal ratio (i.e. the average movement in the irregular component, divided by the average movement in the trend component). While the asymmetric weights enable trend estimates for recent periods to be produced, they result in revisions to the estimates when subsequent observations are available.

Revisions to trend series may arise from:

the availability of subsequent data;
revisions to the underlying data;
identification of and adjustment for extreme values, seasonal breaks and/or trend breaks;
re-estimation of seasonal factors; and
changes to the end weight parameter.

For more information about ABS procedures for deriving trend estimates and an analysis of the advantage of using them over alternative techniques for monitoring trends, see Information Paper: A Guide to Interpreting Time Series - Monitoring Trends (cat. no. 1349.0).

In the Australian Labour Account, quarterly tables are produced in original, seasonally adjusted and trend terms.

For the purpose of deriving the annual average level from quarterly stocks of jobs and employed persons using an arithmetic average, original quarterly series are used.