7112.0 - Selected Agricultural Commodities, Australia, Preliminary, 2005-06  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 26/07/2007   
   Page tools: Print Print Page Print all pages in this productPrint All

TECHNICAL NOTE OLD BASIS ESTIMATES


OLD BASIS ESTIMATES

1 Prior to running the 2005-06 Agricultural Census, the ABS had maintained its own register of agricultural establishments. However, it was increasingly difficult to maintain this list, to the point where it was no longer viable, and users were increasingly questioning the accuracy of some commodity data.

2
The ABS investigated a number of alternatives for maintaining an agricultural business register and discussed these with key users of agriculture statistics. As a result of this, it was agreed that the ABS should move to a new frame sourced from the Australian Taxation Office's Australian Business Register (ABR) for the 2005-06 Agricultural Census.

3
The ABR-based register consists of all businesses contained on the ABR coded to an ‘agricultural’ industry, as well as businesses which have indicated they undertake agricultural activities. All businesses with a turnover of $50,000 or more are required to register on the ABR, and are provided with an Australian Business Number (ABN). Many agricultural businesses with a turnover of less than $50,000 have also chosen to register on the ABR, for example to access subsides that require an ABN.

4
Moving to the ABR-based register required making changes to many of the methodologies used for compiling agriculture statistics. These included changes in the methods used for determining whether farms were 'in-scope' of the collection, and to ways the data was compiled (please see paragraphs 6-9 of the Explanatory Notes for more information about the change in scope and coverage).


IMPLICATIONS FOR USERS

5 The key implication of the move to the new register is that the 2005-06 Agricultural Census data will not be directly comparable with the historical time series of agriculture data. To provide users with a way of comparing ('bridging') the 2005-06 Agriculture Census data with the historical time series, the ABS has prepared two sets of estimates for selected principal commodities (see paragraph 25).

6
'New-basis' estimates have been prepared for 2005-06 using the new ABR-based register and its associated statistical methodologies. These methodologies have been subjected to rigorous analysis and testing, with the resulting 'new basis' estimates representing the start of the new commodity and livestock series.

7
'Old-basis' estimates have also been prepared for a selection of principal 2005-06 agricultural commodities. These estimates are a best judgement approximation to what the 2005-06 estimates may have been if the ABS had continued to use the previous ABS-maintained list of establishments (with its associated statistical methodologies) to conduct the 2005-06 Agricultural Census.


BRIDGING METHODOLOGY

8 The need for a bridging strategy for the Agriculture Census was outlined in Information Paper: Agriculture Census: ABS Views on Content & Procedures, 2005-06 (cat. no. 7103.0) and discussed extensively with key users in March 2006.

9
The key objective of bridging is to produce estimates that reflect what would have been obtained if no changes to the frame and statistical methodology had been made. Comparing the 'old basis' estimates with the historical time series allows the user to get an indication of the 'real world' change. Comparing the 'old basis' estimates with the 'new basis' estimates gives an indication of the impact the change to the frame and statistical methodology has had on estimates.

10
In order to produce 'old-basis' estimates a sample was selected from the units which could be identified and matched between the old and new frames. These common units formed the basis of the 'old-basis' or bridging estimates. In addition to these common units, all units which were common to both frames and did not require checking to reconcile differences between the way the business is represented on the frame and how it operates in the real-world contributed to the bridging estimates. The diagram below illustrates the contribution to bridging estimates by the two types of common units.

Diagram: Bridging Methodology



METHODOLOGICAL CHANGES


11 There are two methodological changes that affect the new-basis estimates: the change in frame, specifically the change in scope, and the units model; and the change in statistical methodology, specifically the change in procedures that reconcile differences between the way the business is represented on the frame and how it operates in the real-world.


(a) The change in frame

12 The change in frame has resulted in the frame size increasing from 150,000 to 190,000 units. Not only were units added to the new frame (that is, became in-scope where they were not in-scope before, see A in Diagram and Table A) but some were removed (became out-of-scope, see C in Diagram and Table A). Changing the size and make-up of the frame will result in change to estimates even if every unit has not changed the data reported the year previously.

13
To gauge the impact of the frame change, the estimated 2005-06 values of the units that were newly out-of-scope (i.e. that had been removed from the 'old-basis' frame - see C) were compared to those of the newly in-scope units (i.e. that had been added to the frame - see A) in the sample.

14
No 2005-06 data was available for sampled units that were newly out-of-scope i.e. row C. Instead, the data these units had reported for the last three years was used to determine the average proportion of 'old-basis' estimates due to these units. Applying this proportion of the 'old basis' estimate from the common units allowed an estimate of the contribution to 'new basis' estimates from the newly out-of-scope units to be produced.


(b) The change in statistical methodology

15 The change in statistical methodology has resulted in a change in the way that some units are treated and how their data contributes to estimates (see B1 in Diagram and Table A). The impact of this change was the most difficult to measure.

16
One of the most important changes has been the way in which holdings owned by a unit were followed up when they were sold or otherwise disposed of by that unit. This process is undertaken to address known discrepancies between the frame and the real world-situation that exist at the time of sample selection. Under the old methodology, each unit in the survey which had sold or otherwise disposed of land would have been investigated to determine details of the new land-holder. If that land-holder was not already on the frame, a survey form would have been dispatched and data gathered. The new methodology dictate that this will only occur for units with an Estimated Value of Agricultural Operations (EVAO) or equivalent of greater than $3 million, to ensure coverage of the largest land holdings does not fluctuate due to changes in ABN registration details of the operating businesses.

17
No 2005-06 data was gathered for the sampled units under the $3 million EVAO cut off, so a value was assigned to reflect the data that would have been obtained under the old procedures. A methodology known as imputation was used to assign this value. The aim of imputation is to assign to a unit with missing data, a value as close as possible to the one that would have been obtained had data been collected. Having no data for these units made imputation difficult and it had to be assumed that these units would be similar in characteristics to the units which had not newly acquired their holding.


DECOMPOSITION

18 For illustration purposes only, table A provides an indication of what the contribution to 2005-06 'old-basis' and 2005-06 'new-basis' estimates of grain sorghum and oats production from the change in frame and change in statistical methodology may have been if the processes outlined above were used. The decomposition consists of four parts:

  • A - the contribution to estimate of units that are in-scope for the new frame but not for the old frame;
  • B1 - the contribution to estimate of units that are found on both frames and did require treatment to reconcile differences between the way the business is represented on the frame, and how it operates in the real world;
  • B2 - the contribution to estimate of units that are found on both frames and did not require any treatment to reconcile differences between the way the business is represented on the frame, and how it operates in the real world; and
  • C - the contribution to estimate of units that are in-scope for the old frame but not for the new frame.

19 Whereas the contribution to 2005-06 'new-basis' estimates could be derived using the above processes, the ABS has quality concerns with the contribution to 2005-06 'old-basis' estimates from continuing units that did require follow up to reconcile frame differences. The contribution from these units has therefore not been included in table A, and consequently the total 2005-06 'old-basis' estimate derived using the processes outlined above are also not included. Readers should note that these totals will differ to those presented in table B as the estimates provided in table B were derived using an alternative strategy as outlined in paragraph 21 below.

Table A, Illustrative bridging decomposition data(a)

Old basis estimate
New basis estimate


2004-05
2005-06
2005-06
% contribution
to 2005-06

Oats-production (t)

A: Units in scope of new frame but not in scope of old frame
na
na
560
32
B1: Continuing units that did require follow up to reconcile frame differences
280
np
408
24
B2: Continuing units that did not require follow up to reconcile frame differences
615
690
755
44
C: Units in scope of old frame but not in scope of new frame
388
512
na
na
Total
1 283
np
1 723
100

Grain sorghum-production (t)

A: Units in scope of new frame but not in scope of old frame
na
na
723
36
B1: Continuing units that did require follow up to reconcile frame differences
621
np
501
25
B2: Continuing units that did not require follow up to reconcile frame differences
819
721
775
39
C: Units in scope of old frame but not in scope of new frame
571
534
na
na
Total
2 011
np
1 999
100

na not available
np not available for publication but included in totals where applicable, unless otherwise indicated
(a) The estimates in this table were derived using the processes outlined in paragraphs 12-17 of the technical note. The 2005-06 'old-basis' estimates will differ to those provided in table B.



FINAL BRIDGED ESTIMATES

20 As a result of these difficulties and resulting quality concerns, the ABS explored an alternate strategy for producing 'old-basis' estimates for 2005-06.

21
Key considerations in the alternate strategy have included:

  • analysis of the 'old-basis' estimates produced using the strategy outlined above;
  • known movements between 2004-05 and 2005-06 for those units common to both the old and new frame;
  • analysis of historical trends in the commodity estimates; and
  • information sourced from relevant industry bodies.

22 'Old-basis' estimates based upon this strategy are presented in tables B and C at the end of this Technical Note.


INTERPRETING THE BRIDGED DATA

23 Users should recognise the preliminary nature of both the 'new-basis' and 'old-basis' 2005-06 estimates. They should also exercise a degree of caution when interpreting the tables as both sets of estimates are subject to variability; the 'old-basis' estimates to both sampling and non-sampling error and the 'new basis' estimates to non-sampling error.

24
The non-sampling variability in the 'old-basis' 2005-06 estimates in particular, is expected to be significant due to difficulties experienced in collecting, processing and estimating 'old basis' data from a 'new-basis' framework. For this reason, the 2005-06 'old-basis' estimates should be used with considerable caution and movements between 2004-05 and 2005-06 considered and indicative only.

25
The following tables allow users to compare 2005 and 2006 estimates subject to the caveats mentioned.

Table B, Production and area of principle crops(a) - Year ended 30 June

Aust. Old Basis(b)
2006 New Basis(c)
2005
2006(d)
Aust.

Barley for grain

Production ('000 t)
7 740
9 480
9 641
Area ('000 ha)
4 646
4 408
4 481

Canola

Production ('000 t)
1 542
1 370
1 454
Area ('000 ha)
1 377
934
996

Cotton lint

Production ('000 t)
563
550
570
Area ('000 ha)
304
318
336

Grain Sorghum

Production ('000 t)
2 011
1 980
1 999
Area ('000 ha)
755
772
792

Lupins for grain

Production ('000 t)
937
1 290
1 357
Area ('000 ha)
845
809
853

Oats for grain

Production ('000 t)
1 283
1 700
1 723
Area ('000 ha)
894
939
945

Rice for grain

Production ('000 t)
339
960
982
Area ('000 ha)
51
98
100

Sugar cane cut for crushing

Production ('000 t)
37 822
37 900
37 990
Area ('000 ha)
434
403
406

Wheat for grain

Production ('000 t)
21 905
24 950
25 704
Area ('000 ha)
13 399
12 337
12 703

(a) Estimates for 2005 are final; estimates for 2006 are preliminary and may be subject to revision. Caution should be exercised in comparing 2005 and 2006 estimates.
(b) Used an ABS-maintained land-based frame (list of 'farms').
(c) Use the Australian Business Register-based frame (list of Agricultural businesses).
(d) Data for 2006 old basis are bridged estimates. Caution should be exercised in comparing 2005 and 2006 estimates.

Table C, Livestock numbers(a) - Year ended 30 June

Aust. Old basis(b)
2006 New basis(c)
2005
2006(d)
Aust.

Milk Cattle(e)

Cows in milk and dry ('000)
2 076
1 865
1 881
Other milk cattle ('000)
981
905
912
Total milk cattle ('000)
(f)3 056
2 770
2 793

Meat cattle

Bulls and bull calves intended for service ('000)
659
687
721
Other calves under one year ('000)
5 357
5 571
5 751
Cows and heifers one year and over ('000)
12 935
12 899
13 456
Other cattle one year and over ('000)
5 776
5 843
6 126
Total meat cattle ('000)
24 725
25 000
26 054

Sheep and lambs

Sheep ('000)
71 947
66 655
67 552
Lambs under one year ('000)
29 178
24 845
25 176
Total sheep and lambs ('000)
101 125
91 500
92 728

Lambing

Ewes mated to produce lambs ('000)
46 147
42 803
43 455
Lambs marked ('000)
37 223
35 117
35 671
Proportion lambs marked to ewes mated (%)
81
82
82
Ewes expected to lamb next year ('000)(g)
46 904
47 292
45 094

Pigs

Boars ('000)
12
12
13
Breeding sows ('000)
286
286
306
Gilts intended for breeding ('000)
43
47
50
All other pigs ('000)
2 197
2 215
2 386
Total pigs ('000)
2 538
2 560
2 755

(a) Estimates for 2005 are final; estimates for 2006 are preliminary and may be subject to revision. Caution should be exercised in comparing 2005 and 2006 estimates.
(b) Used and ABS-maintained land-based frame (list of 'farms').
(c) Used the Australian Business Register-based frame (list of agricultural businesses).
(d) Data for 2006 old basis are bridged estimates. Caution should be exercised in comparing 2005 and 2006 estimates.
(e) Excluding house cows.
(f) See Explanatory notes, paragraph 19.
(g) Forecast made at the beginning of each season.