3302.0.55.001 - Life Tables, States, Territories and Australia, 2011-2013  
Previous ISSUE Released at 11:30 AM (CANBERRA TIME) 06/11/2014   
   Page tools: Print Print Page Print all pages in this productPrint All RSS Feed RSS Bookmark and Share Search this Product
  • Explanatory Notes
  • Technical Note: The Impact of the Revision to Estimated Resident Population on Life Expectancy Estimates (Technical Note)

TECHNICAL NOTE: THE IMPACT OF THE REVISION TO ESTIMATED RESIDENT POPULATION ON LIFE EXPECTANCY ESTIMATES


Introduction

The purpose of this technical note is to measure the impact of revision to Estimated Resident Population (ERP) on life expectancy estimates. This technical note should be read in conjunction with Technical Note 1: The Effect of Rebasing on Estimates of Life Expectancy which was released in Life Tables, States, Territories and Australia, 2010-2012 (cat. no. 3302.0.55.001).

Rebasing and recasting of ERP

After each Census, the ABS calibrates ERP to the latest Census count (this is known as rebasing). As part of this process, the ABS adjusts ERP to take into account the difference between unrebased estimates and the preliminary Census-based estimates by distributing this difference evenly across each quarter over the five-year period since the previous Census.

Following the 2011 Census, the ABS revised down the historical ERP data for the 20-year period from September 1991 to June 2011. This revision process is referred to as ‘recasting’ the data. The decision to recast historical ERP data was made in response to the unusually large difference between the unrebased estimates and the Census-based estimates, resulting from a change in the methodology used to estimate undercount in the 2011 Census. The largest adjustment was required for young adults (15 to 29 years). This age group is traditionally the most difficult to enumerate in the Census, in part due to its mobility. For more information, see Recasting 20 Years of ERP in Australian Demographic Statistics, December Quarter 2012 (cat. no. 3101.0).

Non-revision of life tables

Each year the ABS produces a new set of life tables for Australia, States and Territories. Life expectancy is a summary measure of mortality and therefore changes very slowly. For instance, life expectancy at birth for Australia changed by around 0.24 years per year, from 1901 to 2001. This change in life expectancy was the combined effect of changes in population and mortality data from year to year. Changes in population alone will have minimal impact on life expectancy estimates. For this reason, life tables are never revised, even after the finalisation of individual Census results. This means that life expectancy estimates are not strictly comparable across intercensal periods, though the differences are small and do not limit their effective use as a time series. This is less true for comparisons between the 2010-12 and the earlier life tables as the 2010-2012 life tables used recast ERP and the others did not.

Impact of revision to ERP on life expectancy estimates

At the national level, ERP was revised down by over 293,000 people due to recasting (see Table 1). Of all the states/territories, Queensland experienced the largest downward revision to ERP (106,400 people), hence the largest impact of this revision on life expectancy estimates is expected to be for Queensland. For this reason, two sets of life tables for Queensland for the 2010-2012 period are produced: one set based on the recast ERP and the other on the unrebased ERP. Methods (including graduations) and deaths data used remained exactly the same in both sets. Therefore any difference between the resulting life expectancy estimates will solely be due to the differences in the two underlying ERPs used.

Table 1. SIZE OF THE REVISION TO ERP, States and territories, 2011

Age group (years)
NSW
Vic.
Qld
SA
WA
Tas.
NT
ACT
Aust.(a)

0-4
-5 812
5 987
12 270
2 631
2 628
1 724
203
552
20 142
5-14
3 025
6 879
5 538
401
-546
-51
148
-166
15 228
15-24
38 441
16 776
23 599
2 853
2 997
462
-167
-3 185
81 640
25-34
33 284
15 755
20 259
2 174
-2 077
-1 489
-906
-42
66 529
35-44
9 278
10 343
10 624
1 529
-3 070
-173
-337
-667
27 428
45-54
6 844
11 042
12 038
2 400
-1 352
-473
-25
428
30 899
55-64
6 614
11 341
12 364
3 500
1 724
-35
28
555
36 107
65+
-1 920
4 672
9 664
1 955
1 557
-1 023
652
-120
15 497
All ages
89 754
82 795
106 356
17 443
1 861
-1 058
-404
-2 645
293 470

(a) Includes Other Territories.
The ABS estimated that the use of the recast ERP for Queensland resulted in an estimate of life expectancy at birth around 0.13 years and 0.10 years lower for males and females respectively than the use of the unrebased ERP (see Table 2). This difference was smaller at the older ages; 0.08 years for males and 0.07 years for females at 60, and 0.07 years for males and 0.05 years for females at 70. There was no difference at ages 90 or 100.

Table 2. COMPARISON OF LIFE EXPECTANCY ESTIMATES, Selected ages, Based on the recast and unrebased ERP, Queensland, 2010-2012

Males
Females


Difference:
Difference:
Ages (years)
Recast
ERP
Unrebased
ERP
Recast -
Unrebased
Recast
ERP
Unrebased
ERP
Recast -
Unrebased

0
79.50
79.63
-0.13
83.99
84.09
-0.10
10
70.01
70.14
-0.13
74.46
74.56
-0.10
20
60.20
60.32
-0.12
64.58
64.67
-0.09
30
50.62
50.74
-0.12
54.77
54.87
-0.10
40
41.14
41.25
-0.11
45.03
45.12
-0.09
50
31.85
31.95
-0.10
35.51
35.59
-0.08
60
23.06
23.14
-0.08
26.33
26.40
-0.07
70
15.08
15.15
-0.07
17.69
17.74
-0.05
80
8.49
8.51
-0.02
10.13
10.13
0.00
90
4.24
4.24
0.00
4.81
4.81
0.00
100
2.21
2.21
0.00
2.38
2.38
0.00



The reason the revision to ERP had such a small impact on the life expectancy estimates was that 63% of the Queensland revision was between the ages of 15 to 54 years (see Table 1) where mortality rates are relatively low. Had this revision taken place at the high mortality ages, the impact on life expectancy would have been greater.

Victoria also experienced a large downward revision to ERP (82,800 people) but the revision was smaller than for Queensland. Therefore, the impact of recasting on life expectancy estimates was smaller for Victoria than for Queensland. For example, the male life expectancy at birth estimate for Victoria based on the recast ERP was 0.08 years lower than that based on the unrebased ERP; the comparative figure for Queensland was 0.13 years.

Table 3. COMPARISON OF LIFE EXPECTANCY ESTIMATES, Selected ages, Based on the recast and unrebased ERP, Victoria, 2010-2012

Males
Females


Difference:
Difference:
Ages (years)
Recast
ERP
Unrebased
ERP
Recast -
Unrebased
Recast
ERP
Unrebased
ERP
Recast -
Unrebased

0
80.53
80.61
-0.08
84.55
84.60
-0.05
10
70.90
70.98
-0.08
74.85
74.90
-0.05
20
61.07
61.15
-0.08
64.95
65.00
-0.05
30
51.40
51.48
-0.08
55.10
55.14
-0.04
40
41.83
41.90
-0.07
45.31
45.36
-0.05
50
32.45
32.50
-0.05
35.77
35.81
-0.04
60
23.53
23.58
-0.05
26.54
26.57
-0.03
70
15.41
15.45
-0.04
17.80
17.81
-0.01
80
8.68
8.70
-0.02
10.16
10.16
0.00
90
4.34
4.34
0.00
4.80
4.80
0.00
100
2.28
2.28
0.00
2.36
2.36
0.00



While the impact of recasting on life expectancy has been small, life tables based on recast ERP are not strictly comparable with those based on unrebased ERP. The ABS recommends that they should be used with caution when making comparisons over time, particularly for Queensland.