6203.0 - Labour Force, Australia , Feb 1999  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 25/01/2001   
   Page tools: Print Print Page Print all pages in this productPrint All
  • About this Release
  • Revisions to monthly Labour Force Survey estimates (Feb, 1999) (Feature Article)

Special Article - Revisions to monthly Labour Force Survey estimates (Feb, 1999)


(This article is taken from Labour Force Australia February 1999 ABS Cat. Number 6203.0)

EMBARGO: 11:30 AM (CANBERRA TIME) FRI 26 MAR 1999


INTRODUCTION 

Each year, all seasonally adjusted and trend Labour Force Survey (LFS) estimates produced since February 1978 are revised after the application of updated seasonal factors. Every five years, LFS estimates are revised when updated civilian population estimates become available from the Census of Population and Housing. The latest such revision, incorporating results from the 1996 Census, has resulted in revisions to all original LFS estimates for the period January 1995 to January 1999. Consequently, all seasonally adjusted and trend estimates have also been affected for this period. Changes resulting from both sets of revisions were introduced together in the February 1999 LFS.

This article explains the annual seasonal reanalysis process and the methodology for seasonal and trend series adjustment applied to LFS estimates. It also describes the five-yearly revision of civilian population estimates and examines the impact of these revisions on some of the main estimates produced from the survey.

The impact of the revisions is illustrated in the attachment at the end of this article, where the revised estimates are compared with previously published estimates for selected seasonally adjusted series.


SEASONAL REANALYSIS

Seasonal adjustment is a means of removing the estimated effects of regular seasonal variations from a time series of data, so that the effects of other influences on the series can be recognised more clearly. The seasonally adjusted estimates also reflect the sampling and non-sampling errors present in the original data.

As with all ABS seasonally adjusted series, seasonal factors are reviewed when estimates for the latest year become available. This process assesses possible changes in seasonality, taking into account the most recent year's data, in order to derive the seasonal adjustment factors for the coming year and any revisions to seasonal factors for previous years.


SEASONALLY ADJUSTED SERIES

For Australia, seasonally adjusted estimates for employed persons are obtained by adding the following independently adjusted series for males and females:

  • aged 15 to 19 employed full time;
  • aged 20 and over employed full time; and
  • employed part time.

Original estimates of married females employed full time and employed part time are also independently seasonally adjusted, but are not used in the calculation of total female seasonally adjusted employment.

Seasonally adjusted estimates for unemployed persons are obtained by adding the following independently adjusted series for males and females:
  • aged 15 to 19 looking for full-time work;
  • aged 20 and over looking for full-time work; and
  • looking for part-time work.

Original data for males and females aged 15 to 19 looking for first full-time job are also seasonally adjusted, but are not used in the calculation of the aggregate estimates. The series for married females looking for full-time and part-time work are also independently adjusted, but are not used in the calculation of total female seasonally adjusted unemployment.

Seasonally adjusted estimates for the labour force are formed by adding the seasonally adjusted component series for employed and unemployed. Unemployment rates have been calculated using seasonally adjusted unemployment and labour force estimates. Participation rates have been produced using seasonally adjusted labour force and unadjusted population estimates.

For the States and Territories, seasonally adjusted estimates of employment and unemployment have been obtained by the addition of independently adjusted series for males and females. Estimates of males and females employed full time are also adjusted, but are not used to generate total employment estimates.

Given that estimates for Australia, States and Territories are adjusted independently, the addition of seasonally adjusted estimates across all States and Territories will not generally equal the seasonally adjusted Australian total. Seasonally adjusted estimates for the Northern Territory and the Australian Capital Territory are not published due to the high level of revision to these estimates after each seasonal reanalysis. See paragraph 34 of the Explanatory Notes for further information.

Seasonally adjusted estimates of long-term unemployment are obtained by independently adjusting the original series for males and females unemployed for 52 weeks and under 104 weeks; and the series for males and females unemployed for 104 weeks and over. Total long-term unemployment estimates are the aggregation of these component series.

The seasonal pattern of each ANZSIC Industry Division is analysed and individually adjusted. For this reason, the sum of seasonally adjusted employment estimates for Industry Divisions will not generally equal the seasonally adjusted estimate of total employment. For further information on the seasonal adjustment process, contact the Assistant Director, Time Series Analysis on 02 6252 6076.


TREND SERIES

Smoothing the seasonally adjusted series produces a 'trend' series by reducing the impact of irregular movements in the series. A trend series can be used to analyse the underlying behaviour of the series over time. A trend series is published corresponding to each seasonally adjusted series.

For monthly series, a 13-term Henderson-weighted moving average is applied to the seasonally adjusted series, for all months except the last six. The last six monthly trend estimates are obtained by applying surrogates of the Henderson average to the seasonally adjusted series, and are revised as later data becomes available. For quarterly series, a 7-term Henderson moving average is applied to the seasonally adjusted series, with surrogate weights for the current end of the time series. See paragraphs 32 and 33 of the Explanatory Notes for further information.

Trend series are obtained by applying Henderson weights to each of the component seasonally adjusted series, identified above. Like their seasonally adjusted counterparts, aggregate level trend data are the addition of these component series. Unemployment rates are calculated by using trend unemployment and trend labour force estimates, while participation rates are calculated by using trend labour force data and unadjusted population estimates.


POPULATION BENCHMARKS

LFS estimates of persons employed, unemployed and not in the labour force are calculated in such a way as to add to independent estimates, or 'benchmarks' for the civilian population aged 15 and over.

The benchmarks are based on Census of Population and Housing data, adjusted for under-enumeration and updated monthly for births, deaths, interstate and intrastate migration, and net permanent and long-term overseas migration. Benchmarks are classified by State/Territory of usual residence, part of State of usual residence (Capital city/Balance of State), age and sex. Each cross-classification of these benchmark variables is known as a benchmark cell.

Expansion factors or weights are applied to the survey's sampled respondents to derive estimates that relate to the whole population in the scope of the survey each month. Each sample respondent is allocated a weight depending on their benchmark cell and State/Territory of enumeration.

Labour Force estimates for each characteristic of interest are formed by summing the weights of the persons in the sample with that characteristic. The weighting procedure reduces sampling variability by ensuring that estimates conform to the benchmark distribution of the usually resident civilian population by age, sex and geographic area, while simultaneously compensating and adjusting for any under-enumeration or non-response in the survey.


BENCHMARK REVISIONS

Beginning with the February 1999 survey, LFS estimates will be calculated using population benchmarks based on the 1996 Census of Population and Housing.

LFS estimates have been revised based on updated population benchmarks (incorporating 1996 Census data) for the period January 1995 to January 1999. Potentially, the period affected by the new benchmarks was from February 1994, when the previous revisions took effect. However, an examination of the revised population data showed that the effects on survey estimates prior to January 1995 were too small to warrant revision.


CIVILIAN POPULATION

Over the period January 1995 to January 1999, population benchmarks were revised by no more than 0.1%. The maximum upward revision was 10,900 in February 1996, while the maximum downward revision was 13,300 in September 1997.

REVISION TO POPULATION BENCHMARKS





Male population benchmarks were revised downward for the entire period (to a maximum of 0.2%), falling by a minimum of 2,200 in December 1995 and by a maximum of 13,200 in September 1996. In contrast, the female population benchmarks were predominantly revised upward (again, to a maximum of 0.2%). The female population benchmarks were revised upward by a maximum of 13,500 in February 1996 and revised downward by a maximum of 2,500 in September 1997. The only period of downward revision occurred between June and November 1997.


LABOUR FORCE STATUS

The following graphs compare percentage revisions to seasonally adjusted estimates of employed, unemployed, and not in the labour force, with percentage revisions to the population benchmarks.

PERSONS, PERCENTAGE REVISION







In percentage terms, the revision to population benchmarks and seasonally adjusted employment estimates were closely aligned across the entire revision period. Unemployment estimates were revised downward at a generally increasing rate over the revision period, while estimates of persons not in the labour force were revised upward for the entire period.

Changes in the age distribution of the latest population estimates are a contributing factor to the degree of revision to labour force status estimates. For example, with their generally higher proportion of unemployed (compared to the total population), the downward revisions in the latest estimates of 20-24 year old males and females contributed to the downward revisions to unemployment estimates. Similarly, the generally higher estimates of persons aged 60 to 64 contributed to upward revisions to the number of persons not in the labour force.

MALES, PERCENTAGE REVISION






Likewise, the downward revision in the male population benchmarks led to a reduction in male employment estimates. Reflecting the older population profile of the most recent population estimates, estimates for males not in the labour force increased, whereas unemployment fell by a larger percentage than the male population benchmark revision.

FEMALES, PERCENTAGE REVISION





The increase in female population benchmarks was reflected in the positive change in the employed and not in the labour force categories, while the unemployed category was revised downward.


EMPLOYMENT

Estimates of employed males decreased for most of the revision period, with a maximum reduction of 10,100 in September 1996. In the final month of the revision, January 1999, the estimate decreased by 6,500. In contrast, changes in the estimates of employed females were positive at the beginning and end of the period. A maximum increase of 7,000 was recorded in December 1995 and an increase of 2,000 was recorded in January 1999. Throughout 1997 and 1998 revisions to estimates of employed females were close to zero.

REVISION TO EMPLOYMENT




UNEMPLOYMENT

Estimates of unemployed males and females decreased for the entire revision period. The decrease in estimates of unemployed males ranged between 1,000 in March 1995 and 3,600 in October 1997, with a revision of 3,400 in January 1999. The female unemployment levels were reduced by as little as 400 in January 1995 and up to a maximum of 2,100 in March 1997.

REVISION TO UNEMPLOYMENT








NOT IN THE LABOUR FORCE

Estimates of males not in the labour force generally increased over the revision period, but commenced with a decrease of 1,700 in January 1995. The largest increase in the estimate for males was 3,300 in August 1998, with an increase of 3,000 in January 1999. The estimates of females not in the labour force increased by a maximum of 7,800 in March 1996, a minimum of 100 in September 1997, with a final revision of 3,800 in January 1999.

REVISION TO NOT IN THE LABOUR FORCE








RELEASE OF REVISED DATA

The introduction of revised population estimates was timed to coincide with the annual reanalysis of seasonally adjusted factors. Revised original, seasonally adjusted and trend labour force data were released in the February 1999 issue of Labour Force, Australia, Preliminary (Cat. no. 6202.0) on 11 March 1999. Revised estimates from January 1978 to January 1999 are available on PC-Ausstats, floppy disk or as a special data service. Revised seasonal adjustment factors from February 1978 to January 2000 are also available on floppy disk or as a special data service.


OTHER SERIES

For the LFS, only major National and State/Territory series are subject to seasonal and trend adjustment. Other LFS series can be adjusted on a consultancy basis by contacting the person listed below.


FURTHER INFORMATION

For further information about revised labour force estimates, contact Michael Johnson on 02 6252 6525, or email michael.johnson@abs.gov.au.