Development of a monthly partial turnover indicator

Businesses provide Business Activity Statements (BAS) to the Australian Taxation Office (ATO) as part of their tax obligations. BAS contains useful information including business turnover, wages and capital expenditure. The ATO provides BAS data to the ABS to be used in statistical production processes. These include the formation of homogenous strata in sample designs and forming benchmark totals for survey estimation.

The ABS is looking to leverage off already available administrative data sources to produce more timely insights into the Australian economy. One such area of recent research has been the production of a monthly business output indicator based on BAS turnover. That is, instead of using BAS turnover just to aid in statistical production, we are exploring how to use these data as the primary source of a new monthly economic partial indicator (PI). This indicator will utilise turnover data from businesses which submit BAS turnover to the ATO on a monthly basis.

The ABS has investigated the challenges in producing a monthly PI. One of the key issues is addressing missingness in the monthly data from:

1. Late responses - businesses submit their BAS form too late to be included in the PI for that particular month.
2. Deaths - businesses don't submit a BAS form as they are no longer operating.
3. Changes in submission frequency - this typically involves a business switching between monthly and quarterly submission of BAS data.

It is challenging to address the missingness as it is not known at the time of non-submission which of the above is the cause. Historical imputation is used to address late response (Type 1 above), where the late submitters' data is imputed using the most recent reported BAS data for the unit. A factor is applied to account for changes in the PI between months. This is derived from the change between continuing responding units in the current month and the most recent reported month.

To account for deaths (Type 2), an additional factor is applied to the historical imputation. This factor estimates the likelihood a business is still alive, and hence should have reported BAS. The likelihood is derived from historical data by observing the frequency of non-submitters subsequently submitting in another month. Probabilities will vary depending on the number of consecutive months of non-submission (note some businesses may not submit for several months due to tardy BAS reporting).

Changes in submission frequency (Type 3) are addressed through a coverage adjustment. This is still being investigated, but the current approach is to apply a weight representing the contribution of quarterly submitters, so as to increase the level of the PI to reflect both quarterly and monthly submitters. This weight is derived from the most recent quarter of BAS submission. It is hoped that this will protect against monthly changes in the PI being impacted by changes in submission frequency.

Quality assessment of the PI has been undertaken throughout its development, where comparable ABS survey statistics have served as a reference point. These assessments have been promising, and future refinement of the PI methodology should further improve its performance.

For further information, please contact Tom Davidson at methodology@abs.gov.au.

The ABS Privacy Policy outlines how the ABS will handle any personal information that you provide to us.