5342.0 - Balance of Payments Statistics, Information Paper on Quality , 1996  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 20/02/1996   
   Page tools: Print Print Page Print all pages in this productPrint All  
Contents >> Chapter 3. Analysis of reliability of the statistics >> Are revisions predominantly positive or negative?

3.74. Table 10 shows, for each of the main balance of payments aggregates, counts of positive, negative or zero revisions derived by deducting the initial estimates for each period from the latest available estimates for those periods. Revisions for the months from January 1986 to June 1994 and the quarters from March quarter 1986 to June quarter 1994 have been analysed. (Revisions to annual data are not analysed here because the small number of observations available would be unlikely to yield useful statistics.) For example the first published estimate of merchandise exports (on a shipping date basis) for January 1991 was $3,905 million. In data consistent with the January 1995 issue of 5301.0 that estimate had been revised to $3,969 million. This is counted as one positive revision in Table 10.


10
: DIRECTION OF REVISIONS, INITIAL TO LATEST ESTIMATES - MONTHLY:- JANUARY 1986 TO JUNE 1994 (a) and QUARTERLY:- MARCH QUARTER 1986 TO JUNE QUARTER 1994 (b)
Number of revisions by direction (c)
Aggregate
Negative
Zero
Positive
Total

M O N T H L Y
Current account
Merchandise exports
45
1
56
102
Merchandise imports
54
2
46
102
Balance on merchandise trade
38
0
64
102
Non-merchandise credits
15
0
87
102
Non-merchandise debits
19
1
82
102
Net non-merchandise
62
0
40
102
Services credits
6
0
96
102
Services debits
13
2
87
102
Income credits
37
1
64
102
Income debits
32
1
69
102
Unrequited transfers credits
46
9
47
102
Unrequited transfers debits
11
2
89
102
Balance on goods and services
34
1
67
102

Balance on current account

52

0

50

102

Capital account
Official sector transactions
33
0
69
102
General government transactions
34
0
68
102
Reserve Bank transactions
8
86
8
102

Q U A R T E R L Y
Current account
Merchandise exports
15
0
19
34
Merchandise imports
19
2
13
34
Balance on merchandise trade
13
0
21
34
Services credits
1
0
33
34
Services debits
4
0
30
34
Income credits
12
1
21
34
Income debits
8
0
26
34
Unrequited transfers credits
16
3
15
34
Unrequited transfers debits
1
0
33
34
Balance on goods and services
8
0
26
34

Balance on current account

17

0

17

34

Capital account
Official sector transactions
9
0
25
34
General government transactions
Foreign investment in Australia
8
0
26
34
Australian investment abroad
14
4
16
34
Reserve Bank transactions
Foreign investment in Australia
0
34
0
34
Australian investment abroad
4
29
1
34
Non-official sector transactions
Foreign investment in Australia
6
0
28
34
Australian investment abroad
7
0
27
34

Balance on capital account

9

0

25

34

Balancing item

26

0

8

34

(a) The latest estimate is consistent with the January 1995 issue of 5301.0.

(b)
The latest estimate is consistent with the December quarter 1994 issue of 5302.0.

(c)
Sign convention: A negative direction for revisions to the balance on current account indicates an increase in the deficit. A negative direction for revisions to the other net balances indicates either an increase in the deficit or, in months when surpluses have been recorded, a reduction in the surplus.


3.75. The direction of revision relates to the upward (positive) or downward (negative) adjustment that the initial estimate has undergone to reach the latest estimate (estimates consistent with the January 1995 issue of 5301.0 and the December quarter 1994 issue of 5302.0). In the case of net series, such as balance on merchandise trade, net services, and total general government transactions, a positive revision indicates an increase in a surplus or a decrease in a deficit and a negative revision is a decrease in a surplus or an increase in a deficit.

3.76. Broadly, the information contained in Table 10, which in the absence of bias in the initial estimates would be expected to show approximately equal numbers of positive and negative revisions over the revisions history for any series, complements the information shown in Table 6. Table 10 provides a long-term view of the revisions process from initial to final estimate whereas Table 6 presents a measure of the short-term performance from initial to one-year estimates. While many analysts focus on the expected short-term behaviour of revisions when assessing the significance of initial estimates, it is useful for others to place this short-term performance within a longer frame of reference.

3.77. When comparing data in Table 6 and Table 10 it should not be forgotten that Table 10 shows only counts of revisions (without regard to size) whereas Table 6 shows only monetary values. The different long-term and short-term perspectives should also be kept in mind. For example, one-year revisions may be on a path that is towards the final outcome, but in some cases there may be a tendency for a one-year revision in one direction to work against a longer term tendency for revision in the other direction.

Sign of monthly revisions

3.78. Looking first at the monthly data in Table 10, it can be seen that in the long-term revisions to the merchandise trade items show little bias in terms of direction of revision - this is consistent with the statistics on the one-year revisions experience shown in Table 6. Both services credits and services debits are strongly positively biased in Table 10, while in Table 6, the positive direction of revisions in the short term is not nearly as marked. This tends to suggest that positive revisions continue to be made after the end of the first year. This is borne out by an examination of Graphs A.7 and A.8 in Appendix 1. A similar position is evident for income. Table 10 indicates that income credits revisions tend to be positively biased while revisions to income debits are strongly positively biased (more than twice as many positive as negative revisions). Table 6 confirms this picture with the direction of one-year revisions to income debits being positive but for income credits the one-year revisions are not consistently in one direction. Graphs A.5 and A.6 are consistent with these findings from Table 3.1 and Table 10. In particular, Graph A.5 indicates that significant positive revisions to income credits are made after the end of the first year. Table 6 also indicates no clear direction in one-year revisions to unrequited transfer debits, while Table 10 shows a strong positive bias for longer term revisions to these estimates; both findings are in keeping with Graph A.10.

3.79. The overall effect of revisions to current account items is summarised in revisions to the balance on current account. In Table 6 a negative direction to short-term revisions is indicated, confirmed by Graph A.11 in Appendix 1 which shows an increasing median ratio during the first 12 revisions (reflecting an increase in the size of the deficit on current account). Table 10, however, shows that an equal number of positive and negative revisions were required to initial estimates for the period January 1986 to June 1994 to reach the latest estimate published in the January 1995 issue of 5301.0. This is in line with the initial median estimate in Graph A.11 being close to the final estimate, and the boundaries of the second and third quartiles of initial estimates being roughly evenly spread around the median.

3.80. Monthly estimates of capital transactions of the official sector are made up of FIA and AIA transactions of the general government sector and the Reserve Bank. In Table 6, revisions to Reserve Bank transactions have a zero median value indicating no obvious bias in the direction of one-year revisions; this is borne out by Table 10 which shows positive and negative revisions being evenly balanced. On the other hand, general government sector capital transactions have revisions that are strongly positive in direction in both Table 6 and Table 10. The same picture is conveyed in Graph A.13 of Appendix 1.

Sign of quarterly revisions

3.81. Turning to the counts of revisions to quarterly statistics shown in Table 10, it can be seen that in the merchandise trade items there is a tendency toward positive revisions to exports and negative revisions to imports. Table 6 is consistent for exports but indicates no clear direction for imports revisions (the median value being -$2 million) measured one year after the initial estimates. Graph A.16 provides further explanation for these findings, showing that the initial imports estimate is virtually equal to the latest estimate on average and that revisions are insignificant in value terms. In Table 10 revisions to both services credits and debits are strongly positively biased. This is consistent with Table 6 for debits but not so for credits where Table 6 indicates a slight tendency for one-year revision to be negative (-$11 million). This indicates that positive revisions to services credits must predominate after the end of the first four quarterly revisions. Examination of Graph A.21 in Appendix 1 confirms this and it can be seen that, even after 12 quarterly revision cycles (i.e. after three years), the revised estimate still represents on average less than 95% of the latest estimate.

3.82. Positively biased revisions are also indicated in Table 10 for both credit and debit estimates for income and for unrequited transfers debits. This pattern is consistent with Table 6 for one-year revisions to the debit estimates but not for income credit estimates. Table 6 shows a small negative median revision of -$7 million for income credits. Again, this suggests that significant positive revisions to income credits occur after the first four quarters. Graph A.19 confirms this. As with revisions to monthly estimates, Table 6 indicates a negative direction to one-year revisions to the quarterly balance on current account (i.e. the estimate is revised to a bigger deficit) while Table 10 presents a neutral picture. Graph A.25 indicates how these outcomes can be reconciled: further offsetting revisions, both positive and negative, are made well after the end of the first year.

3.83. Revisions are rarely made to quarterly capital account transactions of the Reserve Bank and there is no evidence of bias in either Table 6 or Table 10. For the general government sector capital transactions, Table 3.5 indicates positive bias for revisions to quarterly estimates classified to FIA and a neutral revisions experience for transactions classified to AIA. This pattern is broadly consistent with the combined picture presented by Table 6 and Graphs A.26 and A.27 in Appendix 1.

3.84. Revisions to quarterly estimates of capital transactions of the non-official sector are strongly positively biased for both FIA and AIA, with nearly five times more positive revisions than negative revisions to the FIA series, and nearly four times more positive revisions for the AIA series. This is consistent with the patterns portrayed in Table 6 and Appendix 1.

3.85. Revisions to quarterly estimates of the balance on capital account are also strongly positively biased (i.e. revisions tend to increase the surplus). This is consistent with the pattern presented by Table 6 and Graph A.32. Statistics are also included for revisions to the balancing item but, as it is residual in nature and reflects the net outcome of revisions to other components of the current and capital accounts, these measures are not examined here.






Previous PageNext Page