Skip to content

Revision policy in Statistics Denmark

At Statistics Denmark data are subject to revisions in the published figures in accordance with the Code of Practice for the European Statistics, Quality Assurance Framework of the European Statistical System (QAF) and the ESS guidelines on revision policy for PEEIs. This implies that the revisions follow common procedures and principles, and that the revision policies for the individual statistics are public.

This document describes the general principles and procedures for revisions of data in Statistics Denmark. The relevant quality declarations contain revision policies for the individual statistics.

Data are subject to revisions in the published figures

  • if errors occur
  • at regular revisions, including revisions as a result of seasonal adjustment
  • at non- regular revisions/major revisions

The revision policy focuses exclusively on how the revisions of provisional data are managed by Statistics Denmark. The procedure for handling errors can be found in Statistics Denmark's dissemination policy.

Principles for revisions

Statistics Denmark has the following principles for revisions of data:

  • The users must have access to information on the reasons for the revisions
  • When publishing provisional data, they are declared as such and ac-companied by information on forthcoming revisions
  • When publishing revised data it must be stated that data are revised. When the revisions are large or unusual, they should be explained.
  • Regular revision analyses is carried out to support the ongoing improvement of the calculations of provisional figures
  • Revisions are regularly discussed with the users

Criteria for revisions

Regular revisions

One of the quality dimensions of statistics is timeliness in order to ensure that the users get the full benefit of the figures. To ensure timely statistics, the production period must be as short as possible. But in some cases it is not possible to publish statistics quickly enough and it is therefore necessary to develop provisional figures. This means that the figures will be revised at least once, and that this will happen regularly in the production cycle.

Seasonal adjustment will in itself also introduce revisions in previously published figures - even if there are no changes in the historical non-seasonally adjusted figures.

Non-regular revisions/major revisions

The non-regular revisions/major revisions mainly occur in the macroeconomic statistics, which are based on a variety of statistics and international manuals that include national accounts, general government statistics statistics, balance of payments and working time accounts. One of the important qualities of these statistics is that they describe the development in the economy over many years. Consequently, new sources and methods are only incorporated in part, because a change of level awaits a major revision. Similarly, new manual rules are only fully incorporated in intervals of several years. In order to ensure consistency between the different statistical systems, major revisions in the national accounts, general government statistics, balance of payments and working time account are performed according to a common plan, which is published as early as possible.

Non-regular revisions may also occur in other statistics as a result of new data sources and methods. At such revisions the users and interested parties should also be notified as early as possible.

Publication of provisional statistics

It must be indicated when a published statistic is provisional. This should be apparent in both text and tables. Likewise, it shall be easy to distinguish provisional figures in the StatBank.

When conducting a revision, the size and direction of the revision shall be described.

When provisional figures are replaced by final figures or by a later version of provisional figures, the replaced data shall be kept in a separate database. This will make it possible for users as well as for employees in Statistics Denmark to conduct revision analyses.

Revision analyses

Provisional figures are only useful for users if they give an adequate description of a given development. Therefore, regular revision analyses must be performed in order to ensure the quality of the provisional figures.

Revision analyses may differ between statistics, but they should all cover the following questions:

  • Size of the revisions - is the calculation of provisional figures optimal?
  • Bias – the revisions should not be biased
  • Efficiency – revisions should be the result of new information about the past.

Revision policy for each individual survey

For surveys from which provisional figures are published, the quality declaration shall include a description of the revision policy. It must contain a description of the regular revisions and possible major revisions. In addition, it must include a description of how the seasonally adjusted series are revised. Finally, it must contain a description of regular revision analyses and an evaluation of their quality.

Discussions with users

Revisions of published figures are regularly discussed in user and contact committees covering the relevant statistics.