1504.0 - Methodological News, Dec 2010  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 09/12/2010   
   Page tools: Print Print Page Print all pages in this productPrint All

Progress with the AIC Review

The Annual Integrated Collection (AIC) is a key data source for measures of Australia's economic performance for each financial year. It replaced the previous ABS annual collection (the Australian Economy Wide Survey) in 2006-07.

In 2009, senior management of PLIES and MIG decided to review the performance of the AIC after the first two years' worth of estimates were produced (in respect of 2006-07 and 2007-08). Charles Aspden, a former Assistant Statistician of the National Accounts Branch, was brought in as a consultant to review the performance of AIC in mid-2009. His review was completed in a whirlwind three weeks; the opening paragraph of his Review report stated that "NAB found substantial differences in the economic picture portrayed by the AIC core and other data sources, most notably the ABS’s quarterly surveys, both in terms of coherence within each of the two years and growth between them".

The desire to improve the accuracy of AIC estimates, plus enhance the AIC's coherence with other data sources such as the Quarterly Business Indicators Survey (QBIS), has been a driving force behind the efforts of the AIC Review Team in the last year or so. The role that MDMD plays is in providing methodological advice to the Review, in particular pursuing the key work identified by the AIC Review Team which wound up in July 2010.

Statistical Services Branch (SSB) within MDMD has been involved in a number of investigations in assisting the Review. In trying to improve the efficiency of the AIC sample (which currently stands at about 20,000 units selected from the Australian Business Register (ABR)), John Preston (SSB) is experimenting with using a modelling approach which uses Business Activity Statement (BAS) data from the Taxation Office (combined with other information from the ABR) to produce unit-level values for each unit on the AIC frame. This method allows for a flexibility of models for the estimation of key measures of financial and economic performance that are output by the AIC, thereby permitting a powerful use of the available auxiliary information.

Edward Szoldra (SSB) has also completed some work confronting wages estimates between AIC and QBIS. This work attempted to shed some light on the discrepancies in the estimates between these two collections. After adjusting for a number of known differences (such as scope and differences in estimation methodology), comparative estimates of annual movement between the AIC and the sum of 4 quarters of QBIS data showed that some industries had discrepancies beyond what is explicable by sampling error. Further investigations indicated some promising (though not definitive) potential reasons for the discrepancies (including issues around differences in annual and quarterly survey reporting).

In addition, more recent work within SSB has been in addressing issues around coherence as indicated in the September 2010 AIC Review Steering Committee meeting. These projects will include SSB working with NAB to ascertain more precisely their quarterly and annual data requirements in the construction of the National Accounts and some work around the effectiveness of the generalised regression (GREG) estimation methodology used in AIC (and whether it is a viable candidate for use in the QBIS). Lastly, SSB will undertake a project which aims at documenting NAB's Supply-Use Table transformations and how AIC and QBIS data flows through the accounts.

For further information regarding these projects, please contact Edward Szoldra on (02) 9268 4214 or edward.szoldra@abs.gov.au.