Update on the implementation of ANZSIC 2006 for sub-annual time series
ANZSIC 2006 (A06) is an update of the current ANZSIC 1993 (A93) industry classification. This new classification will be applied to all sub-annual time series in the ABS from the September quarter 2009. The classification change will affect most ABS time series published on an industry basis. As A06 is a methodological change and is not related to any real world impact, A06 changes to the characteristics of ABS time series should be removed. In this way time series will remain consistent and comparable across time, and will be more useful for analysis and seasonal adjustment purposes.
Consistent A06 time series are being produced by re-estimating historical time series data collected under the old A93 industry classification, with the intention that they will be as close as possible to the new A06 classifications. Impact estimates to measure any remaining discrepancies between these base series and A06 estimates will be produced during a parallel process which will provide overlapping A06 and A93 estimates. Once the base series have been created and the impact estimates produced, the remaining discrepancy can be backcast out so that a single consistent and comparable A06 time series is produced.
To date, base series estimates have been created for all sub-annual collections and these new time series are currently undergoing seasonal reanalysis. Assessment of the backcast estimates will consider the amount of distortion to the seasonally adjusted movements in the base series, so backcasting should not be performed until seasonally adjusted estimates are available.
Backcasting will be run as a process to smooth out any remaining discrepancies between the new A06 estimates and the base series estimate (this difference is referred to as the A06 impact estimate). The Generalized Backcasting Facility (GBF) is a component of the SEASABS suite that can be used to run backcasting automatically across a group of time series, maintaining relevant aggregative relationships between these series and monitoring the effect of the backcasting process on the integrity of our estimates. Backcasting will be run initially using the impact estimate produced during the first parallel period and these first estimates will be used to test production systems and begin quality assurance of the backcast estimates.
The backcasting process will be iterated as updated impact estimates become available during the following parallel periods and if any updates need to be made to the base series estimates. The final backcasting run will be performed on September Quarter (SQ) estimates for a release of all sub-annual A06 estimates in SQ 2009 publications.
For more detail on the implementation of ANZSIC 2006 for sub-annual time series, please contact Nick von Sanden on (02) 6252 5727 or firstname.lastname@example.org.