TECHNICAL NOTE ESTIMATION METHODOLOGY

INTRODUCTION

1 The availability of Business Activity Statement (BAS) data collected by the Australian Taxation Office (ATO) has provided the Australian Bureau of Statistics (ABS) with opportunities to improve the efficiency of collection designs and estimation for its business surveys, while at the same time reducing the reporting burden placed on businesses. Under taxation law, data may be passed by the Commissioner for Taxation to the ABS for specified statistical purposes. Accordingly, turnover and wages information sourced from ATO BAS data were used to improve the accuracy of the 2016-17 industry estimates which were produced using data items collected directly by the ABS from businesses.


ESTIMATION METHODOLOGY

2 The 2016-17 survey continued to use generalised regression estimation, first introduced in the 2006-07 survey. This estimation method enabled maximum use of observed linear relationships between data directly collected from businesses in the survey and auxiliary information. When the auxiliary information is strongly correlated with data items collected in a survey, the generalised regression estimation methodology will improve the accuracy of the estimates. The auxiliary variables used in this survey were turnover and wages sourced from the BAS data of 2,032,255 businesses (including the direct collect sample).


PRODUCING ESTIMATES

3 The following diagram illustrates the ways in which Australian businesses contributed to the estimates in this publication.

DATA STREAMING

4 For the purpose of compiling the estimates in this publication, data for businesses as recorded on the ABS Business Register (ABSBR) contributed via one of three categories (or 'streams') in accordance with significance and collection-related characteristics.

Completely enumerated (CE) stream

5 The CE stream consisted of directly collected survey data for those units recorded on the ABSBR as having employment of at least 300, plus additional economically significant units and units significant to small state estimates.

Generalised regression estimation stream

6 The generalised regression estimation stream comprised directly collected data for those sampled units which were not in the CE stream and had turnover, in aggregate, above the bottom 2.5 percentile of BAS sales for that subdivision, or were identified as employing businesses (based on ATO information).

Business Activity Statement (BAS) stream

7 The BAS stream comprised data for those non-employing businesses whose turnover, in aggregate, was below the bottom 2.5 percentile of BAS sales for that ANZSIC subdivision.

8 Data for the BAS steam was produced using a technique that used BAS turnover to model income from sales of goods and services and BAS non-capitalised purchases to model purchases. The modelling parameters were based on the relationship between BAS data and reported data for small businesses in the direct collect sample over 3 years and were defined at the industry level. Wages and salaries were modelled as 0 since these businesses were non-employers. Employment was based on the business type of (legal) structure.

9 Estimates for each of the selected industries were produced by aggregating the contributing data streams.


STATE AND TERRITORY ESTIMATES

10 For further information on the method used to compile state and territory estimates refer to Technical Note: State and Territory Estimates in this issue.