Data revision policy
Statistical data are generally revised to incorporate new and improved sources of information. The revision of published data is an essential part of producing high quality statistics.
In 2023, Eurostat adopted a general revision policy that says that:
‘Revision of data is a normal process in the production of European statistics to improve their quality. Revisions follow standard, well-established and transparent procedures.’
Eurostat’s revision policy is based on the European Statistics Code of Practice (CoP) which offers guidance on release dates and revisions of statistics, for example for the indicators under principles 6, 8, and 12 of the CoP.
Eurostat has also published guidelines for the European Statistical System for the revision of PEEIs (Principal European Economic Indicators). They present the theoretical and practical aspects of routine, major and non-scheduled revisions, in accordance with the CoP.
Scope and objectives
Eurostat seeks to apply its revision policy to all statistical domains, aiming to:
- ensure that revisions are carried out in a more harmonised way
- provide clear documentation for users
- provide transparent information on the reasons for, and the scheduling of, statistical revisions.
Publishing information on data revisions
Eurostat publishes information on data revisions for each statistical domain in the metadata files under the headings ‘Data revision – policy’ and ‘Data revision – practice’. These metadata can be found attached to folders or individual datasets in Eurostat’s database (look for the symbol) or can be accessed directly from a dataset.
Sometimes this information is also presented in ‘Statistics Explained’ articles or elsewhere on the Eurostat website.
For some statistical domains, Eurostat also analyses revisions when assessing the quality of data.
Types of revisions
European statistics can be revised through:
- routine revisions – when additional information from the providers of the source data becomes available
- major revisions – these are planned and pre-announced to users (including information on the specific characteristics of the revision policy and practice relevant to the statistical domain in question)
- unscheduled revisions – often occurring when errors are detected.
Routine revisions
These are changes in published statistics that are part of the regular data production process, for example planned updates of already published national statistics.
European aggregates are revised when EU countries transmit new statistics to Eurostat, usually according to an agreed schedule.
A seasonal adjustment process can be another reason for a revision. In this case, revisions should follow the requirements specified in the European Statistical System guidelines on seasonal adjustment.
Major revisions
Major revisions of published statistics occur less often. They can be triggered by:
- the availability of new data that is only collected infrequently (every 5 to 10 years), such as a census, input-output tables or labour cost surveys
- an update to the weightings for the base year of an index series (typically every 5 years)
- a change in the concepts, definitions and/or classifications used to produce the data series. For example, when there is a new classification or version of a classification, a change in international statistical standards, or a new EU statistical regulation.
Major revisions can also occur for a combination of these reasons, alongside improvements to the statistical methodology. The overall aim is to ensure the best possible match between published statistics and the phenomena they measure.
Major revisions are usually planned well in advance. Users are informed in good time and warned when they can expect considerable changes to the time series of European statistics. If possible, users are also informed about the likely impact of major revisions.
In the statistical domain of national accounts, benchmark revisions are treated as major revisions. They take place on a regular basis (every 5 years in the EU), to incorporate the results of changes in basic data sources and/or new estimation methods, alongside any changes in standards.
Major revisions of macroeconomic statistics from national accounts and balance of payments are coordinated across domains and across EU countries, according to the harmonised European revision policy for macroeconomic statistics.
In the statistical domain of social surveys, major revisions can be triggered by a change in the way data are collected, by replacing survey information with information from a new register, or by substantially modifying the survey questionnaire.
Unscheduled revisions
These usually occur if corrections are made when a national statistical office or Eurostat detects or is alerted to errors in published statistics. Eurostat’s error management policy describes how Eurostat corrects different types of errors.