TECHNICAL NOTE ESTIMATION METHODOLOGY
INTRODUCTION
1 The estimates in this publication have been produced using a combination of ABS directly collected data and Business Income Tax (BIT) data sourced from the Australian Taxation Office (ATO).
STATISTICAL UNITS DEFINED ON THE ABS BUSINESS REGISTER
2 The directly collected data have been reported by a sample of businesses, as recorded on the ABS Business Register. The ABS uses an economic statistics units model on the ABS Business Register to describe the characteristics of businesses, and the structural relationships between related businesses. Within large and diverse business groups, the units model is used also to define reporting units that can provide data to the ABS at suitable levels of detail. For more information about statistical units refer to paragraphs 6-11 of the Explanatory Notes.
COLLECTION DESIGN
3 The diagram below illustrates the ways in which Australian businesses contribute to the estimates in this publication.
4 The counts of businesses in the diagram refer to the numbers of businesses on the survey 'frame', or list of business units, from which the sample to produce these estimates was selected. During processing of the surveys, units no longer in operation were detected, and the estimates adjusted accordingly.
5 Businesses were only selected to participate in the Retail and Wholesale Industries Survey (i.e. the direct collect sample, or 'Stream D' businesses) if they met a threshold level of activity. Thresholds were set for each ANZSIC class to account for, on average, 90% of total turnover, and no less than 85% of total turnover, as determined by ATO Business Activity Statement data. Businesses eligible for selection in the direct collect sample were selected from the population using stratified random sampling techniques.
6 Very small businesses with annual income less than the threshold were not eligible for selection. Business Income Tax data was used for these units (known as 'Stream T' businesses). At the time of selection Stream T accounted for 55% of retail businesses, but only 8% of expected total income. Similarly, Stream T accounted for 78% of wholesale businesses, but only 10% of expected income. Only broad income and expense items were available from the BIT file.
Employment estimates for Stream T
7 One implication of the use of BIT data in these statistics is that no direct measure of employment is available for those units which contribute to the estimates through the BIT source. This is because the ATO does not collect information about employment numbers. Unlike financial variables, which have a direct relationship to the data available from the BIT files, employment data are not readily amenable to being modelled using the same techniques. Further work therefore has been undertaken in order to devise a suitable modelling process, which is outlined below.
8 In estimating employment for units whose data were sourced from the BIT files, the new methodology took into account:
- whether a business is recorded as paying wages and salaries
- whether a business is an incorporated entity
- whether a business is a sole proprietorship
- for those businesses that are partnerships of individuals, industry averages (derived from the ABS's Labour Force Survey) in order to estimate the number of partners per partnership
- for those businesses that are partnerships of businesses, the assumption that the number of partners per partnership is zero.
9 For each business, an estimate of employee numbers was derived from its value of wages and salaries (if any) using industry averages. For each unincorporated business, these employee numbers were then added to an estimate of its number of working proprietors or partners, to produce an estimate of the total employment of the business. These estimates were then aggregated to the directly collected data to produce the estimates included in this publication.
10 At the time of preparing the estimates for publication, the latest available BIT file related to the 2004-05 reference period. Growth factors were therefore applied to align it with the 2005-06 reference period.
11 In order to minimise the statistical reporting load for very small businesses, the direct collect sample was divided into two streams. A small proportion (14%) of selected businesses was asked to supply detailed income and expense items, while all other businesses were required to complete only broad level income and expense items.
12 Where detailed data were not collected, the data were prorated using businesses with similar characteristics. For detailed income and expense items, this included businesses in Streams D and T. For commodities, it included businesses in Stream T only.
13 Data were then aggregated and adjusted to account for:
- survey non-response
- ABNs that did not match to business income tax data
- other sources of error such as newly established businesses that were not fully represented on the ABS Business Register.
INDUSTRY GROUP AND CLASS ESTIMATES
14 All business units that contributed to the data carry an industry code at the ANZSIC Class (four-digit) level. The sample used for the ABS collected component of these estimates was designed to give broad estimates at the class level. This publication contains estimates at the ANZSIC Subdivision level only, but ANZSIC Group and Class estimates have been released on the ABS web site in the spreadsheets attached to the web page for
Retail and Wholesale Industries, Australia, 2005-06 (cat. no. 8622.0).
STATE AND TERRITORY ESTIMATES
15 The methodology for producing the state and territory estimates separates businesses entities into two groups:
- those businesses identified as operating in only one state or territory ('single state' businesses)
- those businesses identified as operating across more than one state and/or territory ('multi-state' businesses).
16 For larger businesses identified as multi-state on the ABS Business Register, state and territory data were directly collected. For all other businesses, including single state businesses and smaller businesses identified as multi-state, the business's data was allocated to the state of head office.
17 Given multi-state businesses contribute to more than one state, business counts for the individual states and territories sum to more than the total for Australia.
ANZSIC93 AND ANZSIC06
18 The estimates in this publication are based on the Australian and New Zealand Standard Industrial Classification 2006 edition (ANZSIC06). Data in previous issues were based on the 1993 version of the ANZSIC (ANZSIC93). ANZSIC06 was adopted to provide a more contemporary industrial classification system, taking into account issues such as changes in the structure and composition of the economy, changing user demands and compatibility with major international classification standards.
19 For more information, please refer to Australian and New Zealand Standard Industrial Classification (ANZSIC) 2006 (cat. no. 1292.0) which contains more detail on the 2006 industry classification and concordances between the 1993 and 2006 versions.
20 The industry classification change has resulted in a reduction in scope and size of both the Retail Trade and Wholesale Trade Divisions. This has affected the Retail Trade Division to a greater degree than the Wholesale Trade Division. The following table outlines the main changes for the Retail Trade Division. For example, Bread and Cake Retailing and is now distributed across both Manufacturing and Retail Trade.
COMMODITY ESTIMATES
21 Commodity estimates are not included in this publication. They have been released in spreadsheet format on the ABS web page for
Retail and Wholesale Industries, Australia, 2005-06: Commodities (cat. no. 8624.0).
FURTHER INFORMATION
22 For more information, contact William Milne on (03) 9615 7862.
Print Page
Print All
