TECHNICAL NOTE 1: THE EFFECT OF REBASING ON ESTIMATES OF LIFE EXPECTANCY

Rebasing and recasting ERP

1 After each Census, the ABS rebases ERP to the latest Census count. For the 2011 Census, the ABS made a major methodological change to the Census Post Enumeration Survey (PES), to better estimate the level of undercount in Census counts. This new methodology, Automated Data Linking, resulted in a better, but also considerably lower, estimate of undercount. The ABS also estimated that the PES would have estimated lower levels of undercount for previous Censuses, had the methodology been available.

2 As part of the final rebasing process, the ABS therefore undertook an additional process (referred to as ‘recasting’), whereby revisions to historical population estimates were made back to September 1991. For more information on rebasing and recasting please refer to the following two feature articles released in Australian Demographic Statistics, Dec 2012 (cat. no. 3101.0):

• Final Rebasing of Australia's Population Estimates, September Quarter 2006 - June Quarter 2011
• Recasting 20 Years of ERP

Impact on estimates of life expectancy at birth

3 Population estimates, together with deaths data, are a critical input into the method used to derive life tables and life expectancy estimates. Any major downward revision to population estimates will inevitably have an impact on estimates of life expectancy, as the mortality experience in the population references a smaller number of people (or larger number when estimates are revised upward).

4 Unlike most demographic series the ABS produces, the time and resources required to derive life tables are such that they are never revised, even after a rebasing period is concluded. This means that estimates of life expectancy are not strictly comparable across intercensal periods, though the differences are usually so small as to not limit their effective use as a time series. This is less true for comparisons between the 2009-11 and 2010-12 life tables.

5 The following table shows the difference between recent estimates of state and territory life expectancy at birth, and the new 2010-12 estimates, which used rebased and recasted ERP. The largest impact of the rebasing and recasting process on life expectancy at birth estimates is expected to be for Queensland, given it also experienced the greatest relative revision during rebasing. The ABS plans to undertake further analysis to explore the impact of changing denominators (e.g. unrebased, preliminary rebased and final rebased ERP) on life tables and life expectancy estimates. The results of this analysis will be released in Life Tables, States, Territories and Australia, 2011-2013 (cat. no. 3302.0.55.001) in November 2014.

TABLE 1. STATE, TERRITORY AND AUSTRALIA LIFE EXPECTANCY AT BIRTH - 2005-2007 to 2010-2012
	2005-2007	2006-2008	2007-2009	2008-2010	2009-2011	2010-2012
	years	years	years	years	years	years
	MALES
New South Wales	79.1	79.2	79.5	79.6	79.8	79.9
Victoria	79.5	79.6	79.7	80.0	80.3	80.5
Queensland	78.9	78.9	79.1	79.4	79.5	79.5
South Australia	78.8	79.2	79.3	79.4	79.7	79.8
Western Australia	79.2	79.3	79.5	79.7	80.1	80.1
Tasmania	77.7	77.7	77.9	78.0	78.3	78.7
Northern Territory	72.4	72.6	73.3	74.0	74.9	74.7
Australian Capital Territory	80.3	80.1	80.5	80.5	81.0	81.2
Australia	79.0	79.2	79.3	79.5	79.8	79.9
	FEMALES
New South Wales	83.8	83.9	84.3	84.1	84.2	84.2
Victoria	83.8	83.9	84.1	84.3	84.4	84.5
Queensland	83.6	83.7	83.8	83.9	84.1	84.0
South Australia	83.9	83.8	83.9	83.8	84.0	84.2
Western Australia	84.0	84.0	84.1	84.3	84.6	84.8
Tasmania	82.4	82.3	82.2	82.3	82.5	82.6
Northern Territory	78.4	78.4	79.0	79.2	80.5	80.0
Australian Capital Territory	84.0	84.0	84.3	84.7	84.8	85.1
Australia	83.7	83.7	83.9	84.0	84.2	84.3

Further information

6 For more information see Technical Note 2: Improvements to Mortality Estimation at the Highest Ages, 2010-2012, Final Rebasing of Australia's Population Estimates, September Quarter 2006 - June Quarter 2011 and Recasting 20 Years of ERP.