New Editing and Imputation Methodologies for BAS Data
Quality of auxiliary variables (benchmarks) on the business frame will be improved. The Victorian Methodology Unit have recently undertaken a project to improve the methodologies for editing Business Activity Statement (BAS) data and for calculating turnover and wages and salaries benchmarks for units in the quarterly common frames that are in scope of business collections. Work is currently being undertaken to implement these new methodologies. Plan is to introduce the new BAS benchmarks starting with the March 07 Common Frame (CF). Both the new and current benchmarks will thus be available for units on the March 07 CF.
The current benchmarks will continue to be made available until the Monthly Retail Business Survey will move to using the new benchmarks both for stratification and estimation.
Editing Goods and Services Tax (GST) items (e.g. BAS sales) has undergone significant changes. The new editing methodology makes more use of information available on the BAS return to detect anomalies compared to the old methodology (for example, verifying relationships between data items). The option used to report GST items is also used. Data reported is checked to ensure that it is consistent with the option used by the business. Moreover, the old methodology doesn't distinguish between zero and missing values while the new methodology writes a zero for blank returns that have a finalised status. Quality measures from the editing process were also improved.
Due to limited information on the BAS return that can be used to edit Income Tax Withholding (ITW) items (e.g. wages and salaries), the new editing methodology for these items continues to rely heavily on tolerance checks as the current methodology does. The methodology used to set tolerances, however, was improved.
The main changes introduced in the calculation of Australian Business Number (ABN) level benchmarks include:
- nilling of turnover benchmarks for ABNs with cancelled or no GST role or were GST long-term non-remitters
- use of historical data for imputation
- use of median instead of mean of an imputation class for imputing fully non-reporting ABNs
- accepting zeros as valid and not replacing these with a non-nil impute (except for newly birthed units)
- turnover reported by the lead ABN of a GST group is now prorated to members using modelled income, replacing derived size benchmark (DSB) for proration.
Methodology for calculating Type of Activity Unit (TAU) benchmarks have changed significantly. The current methodology models TAU benchmarks using DSB. The new methodology uses ABN level benchmarks of ABS maintained population (ABSMP) ABNs as source of data for TAU benchmarks. A simple aggregation is done in cases where ABN(s) forming the TAU can be identified. For the majority of the remaining TAUs, ABN level benchmarks that are aggregated at enterprise (EN) level are prorated to TAUs formed from the EN. In a small number of complex units, proration is done at enterprise group (EG) level and is prorated to TAUs belonging to the EG.
For further information please contact Elsa Lapiz on (03) 9615 7364.