GENERALISED REGRESSION ESTIMATION FOR ABS BUSINESS SURVEYS
The availability of Business Activity Statement (BAS) data collected by the Australian Taxation Office provides the ABS with opportunities to improve the efficiency of sample design and estimation for it's business surveys. The ABS business surveys currently use two methods of estimation; number-raised and ratio. While ratio estimation allows the use of a single auxiliary variable to improve the precision of the estimates, generalised regression (GREG) estimation allows the use of more than one auxiliary variable, and hence has the potential to be more efficient (i.e. reduce the current sample sizes for the ABS business surveys with no reduction in the accuracy of the estimates) than number-raised and ratio estimation.
The GREG development project will be conducted in four distinct phases.
The first phase was to conduct an initial evaluation of whether GREG estimation has the potential to deliver any reductions in the current sample sizes for the ABS business surveys. An investigation conducted by Methodology Division (MD) on several business surveys found that using GREG estimation with BAS data should result in substantial reductions in the current sample sizes for ABS business surveys.
The second phase is to further develop and refine the generalised regression estimation techniques, and develop a number of 'environment independent' components that will calculate generalised regression estimates and accuracy measures of these estimates (referred to as the GregEst components). These can then be incorporated into existing ABS business processes. At this stage, MD have almost completed the development of methodological techniques for weighting, estimation, winsorization and variance estimation (using a bootstrap replication method) under a generalised regression estimation framework, while the Technical Services Division have commenced the design and construction of the GREG components to perform these methodological techniques.
The third phase is to evaluate the GREG estimation methodology using BAS data and the GregEst components on a number of business surveys, with the aim of developing business cases for using BAS data in estimation. This phase is expected to commence after the construction and testing of the GREG components has been completed.
The fourth phase is to implement the GREG methodology and support the GregEst components into business processes.
For more information, please contact John Preston on (02) 6252 6970.