Caution on interpreting small business statistics

Following the release of Characteristics of Small Business (cat. no. 8127.0) on 28 April, 2004, there have been public comments on the growth of small business between this survey and a similar survey conducted two years earlier.

In the light of these comments the Australian Bureau of Statistics (ABS) cautions that these surveys are not designed to provide the most accurate measures of growth in the number of small businesses.

This is because the figures are based on independent household survey responses (see footnote) and their main objective is to measure the characteristics of small business and how they have changed over time, rather than change in number of small businesses.

The ABS advises that business surveys (as opposed to household surveys) are the best source for small business growth figures.

Characteristics of Small Business showed that:

  • The age profile of small business operators is getting older. In 2003, 33.0% of operators were aged over 50 compared with 30.6% two years earlier.
  • Although more operators are still male (66.8%), the proportion of female small business operators was continuing to increase.
  • The proportion of small business operators born overseas remained largely unchanged.
  • The proportion of small businesses in place five years or longer increased from 51.0% to 53.3%.
  • There was a substantial increase (9.4%) in the number of small businesses using computers, and a stronger increase (19.3%) in the number of small businesses using the Internet.

Due to the interest in small business growth statistics the ABS will give higher priority to updating the business based statistics on small business trends.

Enquiries on the statistics can be direct to Murray Klee on 02 6252 5452.

The Characteristics of Small Business statistics are based on household surveys run approximately every two years. The survey has been conducted in February 1995, February 1997, November 1999, June 2001 and June 2003. The households in these surveys are not common, which increases the sample variability on the estimates of change. The relative standard error (a technical measure of sample variability explained more fully in the publication) on the change between the surveys is about 2%. That is, there is a 95% possibility that the true estimate will be within plus or minus 4 percentage points of the published estimates.

The sample used for these household surveys is the ABS' Monthly Population Survey. One-eighth of this sample is rotated each month to ensure that no individual is included in the Monthly Population Survey for more than eight months. Normally this is to another household in the same neighbourhood. This happened for the 1999 and 2001 surveys. In addition, every five years, following the Population Census, the ABS completely refreshes its sample for the Monthly Population Survey. This last happened in the last few months of 2001 and the early months of 2002. As a consequence, the sample used for the 2003 sample would reflect not only new households, but also new neighbourhoods, making the sample variability larger this time than usual.