1350.0 - Australian Economic Indicators, 1993  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 31/10/1993   
   Page tools: Print Print Page

Feature Article - The Timeliness of Quarterly Income and Expenditure Accounts: An International Comparison

The timely and accurate production of a complex and interrelated set of accounts such as the quarterly Australian National Accounts inevitably involves compromises and trade-offs. A typical trade-off is the use of preliminary, rather than final, data from a particular collection in order to bring out the accounts in a timely manner. This, however, leaves the accounts subject to revision when the final data become available. New information, such as new or infrequent data collections, becoming available and the introduction of improved compilation methodology may also result in the need for revisions. Each country using the United Nations' A System of National Accounts as its framework adopts its own practices to deal with these revisions. Details of Australia's practice are included in Australian National Accounts: Concepts, Sources and Methods (ABS Catalogue No.5216.0).

The following article compares the revision practices of several national statistical agencies, including the ABS. It has been prepared by Philip Smith, the Director of the National Accounts and Environment Division of Statistics Canada. It is published here with the kind permission of Philip Smith and Statistics Canada.

This paper reports the results of an international survey of national income and expenditure accounts release date practices in national statistical bureaux. The survey was conducted by the author in January-March 1993 by means of a questionnaire mailed to statisticians of several countries.

Respondents to the survey were asked on what date their preliminary income and expenditure accounts estimates for each of the four quarters of calendar year 1991 were officially released. They were also asked to indicate the dates on which each of the subsequent four revised sets of estimates were released. To avoid the possibility of unwarranted generalizations from a single year's experience, respondents were asked whether the year 1991 was a typical one from this perspective, or there were special circumstances which affected the release dates in this particular period. Finally, general information was sought on each country's official revision policy.


RELEASE LAGS

Fifteen countries provided the information which was sought, pertaining to release dates in 1991.(Endnote 1) Table 1 displays the average lag for each country calculated as the simple arithmetic mean of the release lags for the four quarters of the year.(Endnote 2) The basic data are reported in Table 2. A slightly different perspective is shown in Table 3, which presents the computed release lags in days.(Endnote 3) It is common for the release lag to vary over the four quarters of the year.

The release lags for the first estimates are of special interest since they show how long the public must wait before getting a first picture of the country's economic performance during the quarter. Statistical bureaux face a trade off here between timeliness and reliability, so it is interesting to observe the various choices they make in this regard. The average release lags for the preliminary estimates are highlighted in Graph 1. Countries fall into five groups:

  • One country, the United States, released its initial estimates within one month of the reference quarter. These estimates were incomplete in that they covered expenditure-based gross domestic product only; initial estimates of corporate profits, net interest, -national income and gross national product lagged by an additional month.
  • Six countries, Australia, Canada, Germany, the Netherlands, Norway and Switzerland, released their first estimates two months after the reference quarter.
  • Three countries, Japan, Sweden and the United Kingdom released their preliminary estimates around the middle of the third month after the reference quarter.(Endnote 4)
  • Four countries, Austria, Denmark, Finland and Italy, released their initial estimates approximately three months after the reference quarter.
  • Finally, one country, New Zealand, released its preliminary estimates about four and a half months after the reference quarter.

Graph 1. AVERAGE RELEASE LAG FOR FIRST ESTIMATES, 1992
Graph 1. AVERAGE RELEASE LAG FOR FIRST ESTIMATES, 1992



The conclusions here, and in the rest of the paper, pertain to income and expenditure based estimates of GDP only. Other components of the System of National Accounts such as the financial flows are ignored. Some countries have more timely estimates of GDP which are based on the measurement of output by industry. These latter measures are not considered in this paper.

The release lags associated with the subsequent revised estimates also vary considerably by country. Once again, the United States stands out as a unique case, having released its second estimates about two months after the reference quarter and its third set of estimates one month after that. The United States is the only country with twelve, rather than four releases of its quarterly national accounts each year. The Netherlands also stands apart from other countries, with eight quarterly releases annually, its second estimates coming out 119 days after the reference quarter. Most countries released their second set of estimates, on average, between 145 and 190 days after the reference quarter. Germany, New Zealand and Sweden differ somewhat, with average release lags of 204, 216 and 210 days respectively for their second estimates, and Austria and Switzerland had the longest such lags at 248 and 289 days.

The lags applying to the third, fourth and fifth estimates, which are also shown in Table 3, differ even more widely and are best considered within the context of general revision policies, in the next section.


REVISION POLICIES

There are many commonalities to the revision policies of the fifteen countries included in the survey. Most impose fairly strict limitations on when and how frequently revisions are permitted, although two, Australia and New Zealand, have almost no restrictions. Most close their accounts after three or four years, allowing no further revisions thereafter except on special, infrequent occasions when historical revisions are carried out. The general pattern in the majority of countries is that the estimates for a particular quarter are open for revision when those for each of the other quarters in that same reference year are first released, and then annually thereafter for a period varying from two to four years.

Countries following this general pattern most closely include Canada, Denmark, Finland, Japan, Norway and Sweden.(Endnote 5) The following paragraphs describe the specific policies of each country, which the reader can also see illustrated for the year 1991 in Table 2.

Canada's revision policy is quite strict. Revisions are permitted for all quarters during the same calendar year. Thereafter, the accounts are open for revision only once per year, at the time of the release of the first quarter estimates in mid-June. They are open for four such annual revisions, after which time they are considered final and are not normally reopened except for historical revisions which typically occur about once per decade. The estimates at constant prices are converted to a new base period at intervals of roughly five years and this transformation is normally accomplished by chain linking, within the usual revision schedule.

In Australia there are few limitations on revisions. The annual revisions are generally most significant at the time of the release of the fourth quarter accounts in March and currently can go back as far as financial year 1985-86. At the time the second quarter estimates are released, annual revisions are normally, although not exclusively, limited to the previous four financial years. The annual revisions at the time when the first and third quarter estimates are released are normally limited to the latest complete financial year. The quarterly estimates are revised in tandem with the annual estimates and, in addition, are open to other revisions each quarter for up to three years.

In New Zealand there are no limitations on revisions, in part because the Department of Statistics has been releasing quarterly national accounts estimates for just a few years.

The United Kingdom revision policy is unique in that it always permits revisions to the quarters of the previous year as well as those of the current year. Annual revisions normally occur in September, when the second quarter estimates and the annual national accounts "Blue Book" are released, and cover the previous four years. At the time of the other three quarterly releases, normally in June, December and March, revisions are allowed in all quarters within the current and previous years, the "current" year being defined as the one in which the most recent Blue Book was published.

In the United States the first estimates, qualified as "advance", are released near the end of the first month after the end of the quarter. The second and third estimates, qualified as "preliminary" and "final", are released near the end of the second and third months respectively. Ordinarily, annual revisions are carried out each July and cover the preceding three years. Comprehensive historical revisions are carried out at about five-year intervals. The revision dates reported by the United States for the 1991 estimates were not fully typical, for two reasons:
  1. the fourth estimates for the first and second quarters of 1991 and the second estimate for the third quarter were delayed by 8 days due to the release of a comprehensive revision and
  2. the fifth estimates for the third and fourth quarters were delayed by approximately 30 days due to a relocation of the bureau.

In Germany the quarterly estimates are open for revision in March, when the fourth quarter estimates are released, and annually thereafter in August or September, at the time of release of the second quarter estimates. They are open for three such annual revisions, after which time they are considered final and are not normally reopened except for historical revisions, which typically occur every 5 to 10 years.

In Japan the estimates for each quarter are open for revision when those for the subsequent quarter are released. Thereafter, the quarterly estimates are open for revision annually at the time of release of the third quarter estimates. They are open for two such annual revisions, after which time they are considered final and are not normally reopened except for historical revisions, which typically occur every 5 to 10 years.

In Denmark the accounts are normally open for revision in the following two quarters. Thereafter, they are normally open for revision only once a year, in January, following the compilation of the annual accounts in October. The estimates are open for three such annual revisions, after which time they are considered final.

In Finland the accounts are open for revision for all quarters during the same calendar year. Thereafter, they are open for revision twice per year, at the time of the release of the first and third quarter estimates. They are open for two years of such revisions, after which time they are considered final and are not normally reopened except in connection with changes in the base year which occur about once every five years.

In Norway the accounts are open for revision for all quarters during the same calendar year. Thereafter, they are open for revision only once during the year, in May, following the release of revised annual estimates. They are open for three such annual revisions, plus two more annual revisions after that which are attributable to the method used for adjusting the quarterly estimates to annual bench-mark totals.

In Sweden the accounts are open for revision for all quarters during the same calendar year. Thereafter, they are open for revision only once during the year, at the time of the release of the annual national accounts in October or November. They are open for two such annual revisions, after which time they are considered final and are not normally reopened except for historical revisions, which typically occur every 5 to 10 years.

In Italy, the revision policy is somewhat analogous to that of the United Kingdom in that, for the first three quarters, it allows revisions to the current year as well as the two previous years, provided the annual totals for those two years remain unchanged. At the time of the first quarter in April, just after the release of the annual revisions normally covering the past three years at the end of March, the quarterly estimates are open for revision for a period equal to n + 2 where n is the number of years for which the annual revisions were just carried. In 1991, the adoption of a new base year for the constant price estimates starting in 1970 has caused some delay in the release of the quarterly estimates which normally occurs three months after the end of the reference period.

The Swiss approach to revisions is unusual in that it draws a distinction between the original, unadjusted estimates and the seasonally adjusted estimates. -Annual revisions to the quarterly GDP estimates normally occur in late November or early December, when the third quarter estimates are released, and cover the previous two years. At the time of the other three quarterly releases the original estimates, without seasonal adjustment, are not open for revision. The seasonally adjusted quarterly estimates are fully open for revision in all quarterly releases.

In Austria, the estimates for the first three quarters are first revised with the release of the fourth quarter. All quarters are open for revision annually thereafter in June or July for three years. Comprehensive historical revisions are carried out at about ten year intervals.

In the Netherlands, a preliminary estimate of GDP growth is released about two months after the reference quarter. Revised, fully-based results (expenditure, production and income) come out about two months after that. When the fully-based results for the fourth quarter come out around the end of April the other quarters of the same year are also open for revision. The annual accounts are revised in August each year, for the previous three years, and the quarterly estimates are adjusted to the new annual levels with the release of the fully-based second quarter estimates.


CONCLUSION

Countries have different approaches with respect to timeliness and revision of the national income and expenditure accounts. In part, this is due to varying attitudes about the trade-off between timeliness and reliability, with some countries, such as the United States, putting a high premium on timeliness of the initial estimates and others, such as New Zealand, putting greater emphasis on their precision. The degree of experience with quarterly national accounting may be another factor accounting for the variance in release timing and revision policies.

The most common pattern is for countries to release their preliminary income and expenditure accounts estimates a little more than two months after the reference quarter, to revise these as the estimates for subsequent quarters of that calendar year are released, and to continue revising them on an annual basis for two to four years thereafter.


ENDNOTES
  1. One other country, Spain, also responded to the survey but since it only started publishing quarterly national accounts estimates in 1992 it was unable to supply the requested information for 1991.
  2. The average lags reported in Table 1 are most interesting for the first and second estimates. Those for the subsequent estimates say more about how often the estimates are revised than they do about release lags.
  3. The release lag for a particular quarter is defined as the number of calender days elapsed between the first day after the quarter and the release day. Thus, if the first quarter estimates were released on April 30, the release lag would be 29 days.
  4. Beginning in 1993, the Central Statistical Office of the United Kingdom has increased the timeliness of its national accounts. A preliminary estimate of GDP, derived from output data, is released three to four weeks after the reference quarter. A second provisional estimate of GDP is released four weeks later, with the main income and expenditure components. A full set of national accounts becomes available twelve weeks after the reference quarter.
  5. Information provided by Spain, relating to calendar year 1992, indicates that it too follows the general pattern just described. Spain allows revisions for all quarters in the same calendar year. Thereafter, revisions are permitted once a year at the end of June, with the release of the first quarter estimates. The accounts are closed after three annual revisions.


APPENDIX

Table 1. NATIONAL INCOME AND EXPENDITURE ACCOUNTS AVERAGE RELEASE LAGS FOR 1991 BY COUNTRY

Estimate
1st
2nd
3rd
4th
5th

Country
Days

Australia
61
154
245
335
427
Austria
86
248
407
772
1,092
Canada
64
156
322
555
857
Denmark
96
187
307
614
974
Finland
91
184
298
439
599
Germany
65
204
387
570
799
Italy
106
194
286
376
465
Japan
75
166
384
749
977
Netherlands
60
119
256
483
802
New Zealand
130
216
315
406
497
Norway
62
146
299
519
816
Sweden
82
210
406
546
733
Switzerland
62
289
658
936
..
United Kingdom
82
173
265
356
446
United States
27
60
85
229
548

Table 2. NATIONAL INCOME AND EXPENDITURE ACCOUNTS RELEASE DATES FOR 1991 BY COUNTRY

Estimate
1st
2nd
3rd
4th
5th

Country
Quarter
Date

Australia
1st
30-05-91
15-08-91
05-12-91
17-03-92
02-06-92
2nd
15-08-91
05-12-91
17-03-92
02-06-93
13-08-92
3rd
05-12-91
17-03-92
02-06-92
13-08-92
01-12-92
4th
17-03-92
02-06-92
13-08-92
01-12-92
18-03-93

Austria
1st
29-06-91
28-03-92
27-06-92
28-06-93
28-06-94
2nd
28-09-91
28-03-92
27-06-92
28-06-93
28-06-94
3rd
20-12-92
28-03-92
27-06-92
28-06-93
28-06-94
4th
28-03-92
27-06-92
28-06-93
28-06-94
..

Canada
1st
20-06-91
30-08-91
29-11-91
28-02-92
22-06-92
2nd
30-08-91
29-11-91
28-02-92
22-06-92
21-06-93
3rd
29-11-91
28-02-92
22-06-92
21-06-93
20-06-94
4th
28-02-92
22-06-92
21-06-93
20-06-94
19-06-95

Denmark
1st
02-07-91
30-09-91
21-01-92
22-01-93
15-01-94
2nd
30-09-91
21-01-92
30-03-92
22-01-93
15-01-94
3rd
21-01-92
30-03-92
30-06-92
22-01-93
15-01-94
4th
30-03-92
30-06-92
22-01-93
15-01-94
15-01-94

Finland
1st
03-07-91
30-09-91
31-12-91
30-03-92
10-07-92
2nd
30-09-91
31-12-91
30-03-92
10-07-92
29-12-92
3rd
31-12-91
30-03-92
10-07-92
29-12-92
15-07-93
4th
30-03-92
10-07-92
29-12-92
15-07-93
29-12-93

Germany
1st
04-06-91
30-08-91
13-03-92
03-09-92
09-03-93
2nd
30-08-91
13-03-92
03-09-92
09-03-93
07-09-93
3rd
05-12-91
13-03-92
03-09-92
09-03-93
07-09-93
4th
13-03-92
03-09-92
09-03-93
07-09-93
07-09-94

Italy
1st
20-07-91
09-10-91
15-01-92
17-04-92
08-07-92
2nd
09-10-91
15-01-92
17-04-92
08-07-92
08-10-92
3rd
15-01-92
17-04-92
08-07-92
08-10-92
12-01-83
4th
17-04-92
08-07-92
08-10-92
12-01-93
08-04-93

Japan
1st
18-06-91
19-09-91
04-12-91
03-12-92
03-12-93
2nd
19-09-91
04-12-91
03-12-92
03-12-93
..
3rd
04-12-91
19-03-92
03-12-92
03-12-93
..
4th
19-03-92
16-06-92
03-12-92
03-12-93
..

Netherlands
1st
30-05-91
25-07-91
30-10-91
27-04-92
27-10-92
2nd
09-09-91
30-10-91
27-04-92
27-10-92
26-10-93
3rd
25-11-91
30-01-92
27-04-92
27-10-92
26-10-93
4th
27-02-92
27-04-92
27-10-92
26-10-93
25-10-94

New Zealand
1st
20-08-91
30-10-91
19-02-92
28-04-92
28-07-92
2nd
30-10-91
19-02-92
28-04-92
28-07-92
30-11-92
3rd
19-02-92
28-04-92
28-07-92
30-11-92
16-02-93
4th
28-04-92
28-07-92
30-11-92
16-02-93
28-04-93

Norway
1st
10-06-91
09-09-91
09-12-91
10-02-92
11-05-92
2nd
09-09-91
09-12-91
10-02-92
11-05-92
10-05-93
3rd
09-12-91
10-02-92
11-05-92
10-05-93
10-05-94
4th
10-02-92
11-05-92
10-05-93
10-05-94
10-05-95

Sweden
1st
25-06-91
24-09-91
16-12-91
23-03-91
18-11-92
2nd
24-09-91
16-12-91
23-03-92
18-11-92
15-11-93
3rd
18-12-91
23-03-92
18-11-92
15-11-93
..
4th
23-03-92
18-11-92
15-11-93
..
..

Switzerland
1st
04-06-91
29-11-91
01-12-92
07-12-93
..
2nd
30-08-91
29-11-91
01-12-92
07-12-93
..
3rd
29-11-91
01-12-92
07-12-93
..
..
4th
06-03-92
01-12-92
07-12-93
..
..

United Kingdom
1st
24-06-91
23-09-91
20-12-91
19-03-92
24-06-92
2nd
23-09-91
20-12-91
19-03-92
24-06-92
23-09-92
3rd
20-12-91
19-03-92
24-06-92
23-09-92
21-12-92
4th
19-03-92
24-06-92
23-09-92
21-12-92
12-03-93

United States
1st
26-04-91
29-05-91
26-06-91
04-12-91
30-07-92
2nd
28-07-91
28-08-91
26-09-91
04-12-91
30-07-92
3rd
29-10-91
04-12-91
20-12-91
30-07-92
01-09-93
4th
29-01-92
28-02-92
26-03-92
30-07-92
01-09-93


Table 3. NATIONAL INCOME AND EXPENDITURE ACCOUNTS RELEASE LAGS FOR 1991 BY COUNTRY

Estimate
1st
2nd
3rd
4th
5th

Country
Quarter
Date

Australia
1st
59
136
248
351
428
2nd
45
157
260
337
409
3rd
65
168
245
317
427
4th
76
153
225
335
442

Austria
1st
89
362
453
818
1,183
2nd
89
271
362
727
1,092
3rd
80
179
270
635
1,000
4th
87
178
544
909
..

Canada
1st
80
151
242
333
448
2nd
60
151
242
357
721
3rd
59
150
265
629
993
4th
58
173
537
901
1,265

Denmark
1st
92
182
295
662
1,020
2nd
91
204
273
571
929
3rd
112
181
273
479
837
4th
89
181
387
745
1,110

Finland
1st
93
182
274
364
466
2nd
91
183
273
375
547
3rd
91
181
283
455
653
4th
89
191
363
561
728

Germany
1st
64
151
347
521
708
2nd
60
256
430
617
799
3rd
65
164
338
525
707
4th
72
246
433
615
980

Italy
1st
110
191
289
382
464
2nd
100
198
291
373
465
3rd
106
199
281
373
469
4th
107
189
281
377
463

Japan
1st
78
171
247
612
977
2nd
80
156
521
886
..
3rd
64
170
429
794
..
4th
78
167
337
702
..

Netherlands
1st
59
115
212
392
575
2nd
70
121
301
484
848
3rd
55
121
209
392
758
4th
57
117
300
664
1,029

New Zealand
1st
141
212
324
393
484
2nd
121
233
302
393
518
3rd
141
210
301
426
504
4th
118
209
334
412
483

Norway
1st
70
161
252
315
406
2nd
70
161
224
315
679
3rd
69
132
223
587
952
4th
40
131
495
860
1,225

Sweden
1st
85
176
259
357
597
2nd
85
168
266
506
868
3rd
76
174
414
776
..
4th
82
322
684
..
..

Switzerland
1st
64
242
610
981
..
2nd
60
151
519
890
..
3rd
59
427
798
..
..
4th
65
335
706
..
..

United Kingdom
1st
84
175
263
353
450
2nd
84
172
262
359
450
3rd
80
170
267
358
447
4th
78
175
266
355
436

United States
1st
25
58
86
247
486
2nd
25
58
87
156
395
3rd
28
64
80
303
701
4th
28
58
85
211
609