Revision policy of National accounts

Annual results

The publication and revision policy of results is based on the recommendations of the « Harmonised European Revision Policy (HERP)» adopted by the « Committee on Monetary, Financial and Balance of Payments Statistics (CMFB)» in 2012.

Estimates for main aggregates from national accounts for reference year A-1 are published for the 1st time for 7^th or 8^th of March in year A. These annual estimates are calculated based on quarterly national accounts by summing the values of the four quarters of reference year A-1. These estimates are flagged as preliminary as they are still based largely on estimation and extrapolation methods rather than on source data.

Following the HERP recommendations, STATEC applies two types of revision to national accounts:

• the routine revision
• the benchmark revision

The mission of the CMFB is to assist the EU Commission in the elaboration and implementation of the work program concerning monetary, financial statistics and for the balance of payments. CMFB is a forum to ease coordination between representatives from European central banks, national statistical institutes, European Central Bank and Eurostat.

Routine revision

A routine revision spans over a period of the four most recent reference years. Based on annual national accounts, results for reference years A-1, A-2, A-3 and A-4 are revised for the 8^th of September of year A. Results for reference year A-1 keep the preliminary status due to the large share of estimates necessary for obtaining the results.  Results for reference year A-2 are flagged as provisional. Compared to results of A-1, more source data has been integrated, but the share of estimates remains significant. Past experiences show as well that these results may still contain some imperfections like figures which retrospectively need a different interpretation and recording. Results for reference year A-3 are flagged as semi-final as most imperfections are corrected. Results for reference year A-4 are flagged as final as they are not revised any more until the next benchmark revision.

Reference years published with a routine revision

Revision of annual results

Annual national accounts provide annual values for national accounts aggregates whereas quarterly national accounts provides quarterly values for national accounts aggregates, which may be summed up to annual values if all quarters are complete. Quarterly national accounts main use is the provision of estimates for the most recent quarters and as many data sources are still incomplete or missing they use extensively estimation methods. Annual national accounts use a larger range of data sources then quarterly national accounts and are built on calculation methods adapted to these data sources.

To avoid break in statistical series, routine revisions are based in principle on unchanged data sources and calculation methods. Sometimes exceptions are needed for instance when a data source disappears and needs to be replaced with a new one, in case of late discovery of an error in a calculation method or for other minor changes not causing a significant break in series. Avoidance of breaks in statistical series is the guiding principle when deciding about changes to introduce in routine revisions.

Benchmark revision

While a routine revision spans over a period of four reference years, a benchmark revision spans over of longer period often a far longer period.  No fixed rule exists for the depth of a benchmark revision. Currently benchmark revisions start with the first year of the mandatory transmission program of national accounts results to Eurostat.

A benchmark revision is the moment for integrating changes in data sources and calculation methods, which would have caused break in statistical series during a routine revision. When introduced for a benchmark revision, these changes are applied to the whole period covered by the benchmark revision avoiding any break in statistical series. A benchmark revision is as well the moment to introduce changes resulting from the introduction of an update of the methodological reference manuals like the System of National Accounts from the United Nations. Such an update may require new, deletion or updating of calculation methods and or data sources. Results from a benchmark revision are in principle introduced for the 30^th of September. With the HERP, European Union Member States have committed to perform benchmark revision for their macroeconomic statistics at least every five year. Performing a benchmark revision is very resource consuming and few countries are able to perform such revisions each year. Users prefer often as well stable statistical series.

Year of performing a benchmark revision

Revision of annual results

Quarterly results

The first estimate for a certain quarter is published 2 month + 8 days after the reference quarter.

Quarterly national accounts are never revising annual values of aggregates after they have been estimated for the 1^st time. Only annual national accounts are revising annual values of aggregates as calculation methods and data source of annual national accounts are considered more complete and robust. Quarterly national accounts are only revising the quarterly profile, seasonal and calendar day adjusted and non-adjusted, of quarterly results once the corresponding annual values of the aggregates are known. These revisions are performed over the whole period covered by national account results. Revisions for the most distant quarters are of course minor.

Revision of quarterly results

* where n depends on period covered by national accounts