Supplementary Survey methodology changes in the Labour Force Survey

The ABS will be completing the Labour Force Modernisation project at the end of September 2026. From this time, all Monthly Population Survey (MPS) respondents will have an improved digital experience, and users will have access to more frequent insights on selected topics. 

Under the previous Labour Force Survey model, the ABS asked respondents supplementary questions each February and August, which influenced who and how people responded in these months. This influence, referred to as the Supplementary Survey effect, was evident in the original Labour Force data and was accounted for in the calculation of seasonally adjusted and trend estimates to ensure the accuracy of the core labour force estimates. 

As part of Labour Force Modernisation, we have introduced a new survey collection model for the supplementary content, where respondents will receive these additional questions only after completing the core Labour Force questions in their last month (the outgoing rotation group). This model enables greater flexibility to make changes to supplementary content as user requirements evolve, without any influence on core Labour Force questions. Now that the new survey model is in place and answers to core questions cannot be affected by supplementary questions, the adjustment for the Supplementary Survey effect previously seen in February and August is no longer required. 

It is possible that the Supplementary Survey effect adjustment was inadvertently also capturing some seasonality. If this was the case, removing the adjustments for the Supplementary Survey effect could have an unintended impact. Sensitivity testing and external expert advice indicate that any impact will be smaller than the normal month-to-month variation that we already observe due to the Labour Force survey being a sample survey. 

The ABS will continue to undertake extensive quality assurance checks each month to ensure estimates remain fit-for-purpose. The ABS recommends taking this information into consideration in conjunction with trend estimates when interpreting month-to-month changes in the seasonally adjusted estimates for analysis involving February and August data (and surrounding months). 

This article explains the history of changes to the supplementary content and the influence of this change on the current and upcoming Labour Force statistics. 

Prior to 1994, all MPS respondents received a Supplementary Survey in most months, with different topics assigned to specific months of the year. 

From 1994, the Supplementary Surveys were asked in most months to 7 out of the 8 rotation groups. The outgoing rotation group was instead asked a different set of questions, known as the Multipurpose Household Survey. 

From 2014, the frequency of Supplementary Survey data collection was reduced to three months:

• Participation, Job Search and Mobility (PJSM), conducted in February 
• Survey of Education and Work (SEW), conducted in May 
• Characteristics of Employment (CoE), conducted in August 

The supplementary content was still asked of 7 of the 8 rotation groups and was embedded into the survey. That is, labour force and supplementary content was collected for one person prior to moving to the next person in the household.

This change in 2014 introduced a new effect into the data in February and August that was not a reflection of real-world labour market seasonality. The inclusion of additional supplementary content in February and August influenced respondent behaviour, which significantly altered the Labour Force estimates, including numbers of employed and unemployed persons. This impact is referred to as the Supplementary Survey effect. Since 2014, the ABS has removed this effect prior to calculating seasonally adjusted and trend estimates through a process known as prior adjustments. There was no effect observed in the Labour Force data for May months as the additional questions were unrelated to labour force status. 

From August 2026, the supplementary content will be asked each month from the out-going rotation group only (1/8th of the sample) after completing all labour force content from the household. Respondents who do not subsequently complete the Supplementary Survey content will still be included in Labour Force estimates. This new approach is designed to remove the Supplementary Survey effect observed from the old model and provide a more robust model moving forward. The diagram below visualises this change. 

Given this change, the current Supplementary Survey prior adjustments moving forward for February and August are no longer required. The below sections outline how the ABS has removed the Supplementary Survey effect, and ongoing quality assurance of the seasonal adjustment processes used in producing Labour Force estimates.

Since 2014, the ABS has removed the Supplementary Survey effect through a process known as prior adjustments. This process removes non-seasonal factors before the calculation of the seasonally adjusted and trend estimates. This process is also used to remove non-seasonal effects such as outliers, moving holidays (i.e. Easter) and trading day effects. Adjustments are estimated using a regARIMA model (regression with ARIMA based errors). Under this process, regressors indicate when data deviates from historical trends, and the size of the adjustment needed to restore broad historical patterns is calculated using coefficients. The effect is then removed prior to the calculation of seasonal and trend estimates.

The ABS does not expect any Supplementary Survey effects in the original data from August 2026 onwards under the new Labour Force Survey model. As such, prior corrections for the Supplementary Survey effect are no longer required for future February and August data. The regressor factors will be set to zero for all months from August 2026 onwards, and we will continue to apply the regressors from 2014 to February 2026 inclusive.

It is possible that the Supplementary Survey effect adjustment was inadvertently also capturing some seasonality. If this was the case, removing the adjustments for the Supplementary Survey effect could have an unintended impact. The ABS has undertaken sensitivity analysis modelling a range of scenarios to estimate potential month-to-month variation from removing prior corrections and has sought external expert advice on our approach. We expect if there is any impact from this change that it will be smaller than the normal month-to-month variation that we already observe due to the Labour Force survey being a sample survey. 

Additionally, if there is any residual influence from the Supplementary Survey effect in our seasonal adjustment calculations, our standard time series decomposition methods will gradually address this through the revision process over the coming few years, as additional data becomes available.

To ensure that Labour Force estimates remain fit-for-use, additional quality assurance checks have also been implemented, with a strong focus on time series diagnostics. This includes changes in seasonal patterns, outliers, and comparisons with forecasted estimates.   

For further information on how the ABS maintains the high quality of its Labour Force statistics please refer to the article: How the ABS quality assures Labour Force data during times of change. 

As we have done during the Labour Force Modernisation transition, the February 2027 release will be delayed by one week to provide additional time for quality assurance, to ensure the estimates are fit-for-use. 

The ABS does not expect that there will be a Supplementary Survey effect under the new collection model. We will continue to analyse the outgoing rotation group who receive the supplementary questions over the coming months to confirm this expectation. 

If you have any further questions about this article, please email labour.statistics@abs.gov.au.