1216.0.55.003 - Australian Statistical Geography Standard: Design of the Statistical Areas Level 4, Capital Cities and Statistical Areas Level 3, May 2010  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 21/05/2010  First Issue
   Page tools: Print Print Page Print all pages in this productPrint All

STATISTICAL AREAS LEVEL 4 (SA4S)

The SA4 level represents the largest sub-state regionalisation of the main structure in the ASGS. Its main purpose is to provide the geographical basis for labour force statistics.

The current LFRs have a number of weaknesses, which have severely compromised their usefulness.

      1. They are not integrated into the ASGC. They are built from ASGC regions at the time of each Census. The underlying ASGC units are reviewed annually, while the LFRs remain stable. Consequently, between Censuses, the LFRs and the ASGC drift out of alignment, limiting direct comparison until they are realigned at the next Census.
      2. The existing regions are built around administrative boundaries rather than any objective analysis of the Australian labour market geography. In many cases the existing LFRs split labour markets.
      3. Their population sizes are extremely inconsistent; this means the very small regions have high relative standard errors due to the low sample, which is standardised, only within each state.
      4. Many of the more remote LFRs are spatially very diverse although this is to a degree inevitable due to the low population densities.

The ABS will address these weaknesses in the design of the SA4s.

The SA4s will be fully integrated into the ASGS. They will remain stable between Censuses, as will the underlying geographical units. This will allow a direct comparison between Labour Force data and other data released by the ABS on the ASGS main structure.

Labour markets will be a key consideration in the design of SA4s. Labour markets are geographic regions, which have a high degree of interconnectedness or overlap between the labour supply (where people live) and demand (where people work). These clusters of labour supply and demand occur because of the place of work being restricted by the commuting distance. For example, most people who live in Bendigo will work in Bendigo rather than Melbourne. Consequently, labour markets may be identified using travel to work data and the resulting regions provide an ideal platform for the analysis of labour force data.

The ABS has consulted with a number of experts on labour market geography to identify labour markets within Australia through the an analysis of the 2006 Census travel to work data (James Newell, Monitoring Evaluation Research Associates (MERA), New Zealand, Bill Mitchell, Martin Watts and Michael Flanagan, Centre of Full Employment and Equity (CofFEE), The University of Newcastle).

The ABS has worked with James Newell (MERA) to identify Australian labour markets using a version of the Coombes Algorithm and the 2006 Census Statistical Local Area (SLA) travel to work matrix. The resulting labour markets are characterised by a large number of very small regional labour markets comprised of single SLAs with a smaller number of medium sized labour markets around regional centres and very large labour markets representing the major metropolitan centres. While this may be an accurate reflection of Australian labour markets, it does not create an ideal output geography for labour force data which is collected using a sample survey requiring large and consistently sized geographic regions.

CofFEE have produced an alternative labour market geography called CofFEE Functional Economic Areas (CFEAs). These are defined using the Intramax Algorithm (Mitchell, Bill and Watts, 2007) and the 2006 Census SLA travel to work matrix and are available through their website http://e1.newcastle.edu.au/coffee/functional_regions/. The Intramax algorithm uses an incremental aggregation approach to create a hierarchical diagram of commuting relationships between all SLAs, this allows much larger and more consistently sized geographic regions to be created while still respecting the commuting interactions.

The ABS has used these two versions of Australian labour markets to inform the design of the SA4 regions. This process essentially involves amalgamating very small regional labour markets and splitting major metropolitan labour markets into more evenly sized regions whilst preserving the suitably sized regional centre labour markets. Given that metropolitan labour markets are naturally larger than those in regional areas, larger population SA4s were created within the major cities while smaller SA4s were created in regional areas. The desired minimum population of 100,000 was set as a compromise between preserving the labour markets of large regional centres and minimising the extent to which unacceptably high Relative Standard Errors (RSEs) occur. This does not guarantee that all SA4 regions will have acceptable RSEs for all data items. Such a guarantee is impossible given that the sample is designed for the release of data at a state and national level.

State boundaries are a limiting factor in the extent to which SA4s can reflect labour markets. Several labour markets cross state boundaries, for example, Gold Coast-Tweed, and Canberra-Queanbeyan. As far as possible the SA4 design will represent this reality, but the fact that the SA4s must add up to the various states and territories will compromise the representation of some labour markets at the state borders.

In the light of the above:
      1. SA4s will be made up of whole SA2s.
      2. The SA2s will be combined into SA4s based on an analysis of 2006 Census travel to work data.
      3. SA4s will be designed independently of the existing LFRs, although in some areas the results may be similar.
      4. In regional areas, SA4s will represent a single, or clusters of labour markets, with an average population of between 100,000 and 300,000 people.
      5. Capital city labour markets will be broken up into sub markets of between approximately 150,000 and 500,000 thousand people based on an analysis of travel to work data.