13 / 20 Germany: WZ 2025
What are double reporting and backcasting in Germany?
Published Updated
Statistical classification only
Statistical classification only. This guide does not determine any reader's code. Confirm a code with the Federal Statistical Office (Destatis) or the register that holds the record. This is general information, not advice on any reader's registration.
Double reporting and backcasting are the two methods that keep German statistics comparable across the change from WZ 2008 to WZ 2025. Double reporting means sending the same statistics under both the old and the new classification in parallel for a period. Backcasting means re-presenting historical statistics under the new classification, so that a series compiled under the old version can be read alongside a series compiled under the new version. Destatis describes both methods in its WISTA article on the introduction of WZ 2025.
Eurostat states that plans for backcasting and double reporting have been established with its partners in the European Statistical System. The methods serve statistics. They do not change any register entry, and they do not decide any question for a specific business.
How does double reporting work?
Double reporting works by delivering data twice, once under each version, for the same reference period. Destatis states that double reporting for structural business statistics should take place exclusively for the reporting year 2025 (FIXED), under the implementing regulation for European business statistics. For short term business statistics, Destatis states that data for the reporting period from 2025 to 2027 (FIXED range) are to be transmitted twice, and that short term statistics are to be delivered under the old version until the end of 2027 (FIXED), then retrospectively under the new version at the start of 2028 (FIXED), with the change to the new base year. The parallel delivery lets users see the effect of the classification change on the same period, without guessing. It also lets Eurostat check that the two deliveries are consistent, apart from the classification effect itself.
| Method | What is delivered | Period stated by Destatis |
|---|---|---|
| Double reporting, structural business statistics | Data under both versions | Reporting year 2025 (FIXED) only |
| Double reporting, short term business statistics | Data under both versions | Reporting period 2025 to 2027 (FIXED range) |
| Backcasting | Historical data re-presented under the new version | As required by the implementing regulation for the domain |
The table above follows the Destatis description. It states no backcasting period of its own, because backcasting periods are set domain by domain in the implementing regulations, and a reader should check the regulation for the domain concerned.
How does backcasting work?
Backcasting works by applying the correspondence between the old and new classifications, together with source data and methods, to historical figures, so that they can be presented as if they had been compiled under the new classification. The correspondence table shows which headings link. It does not show proportions where a heading splits or merges, so backcasting needs more than the table: it needs the underlying data at a finer level, or a method for allocating the old heading between new headings. Destatis notes that binding requirements for backcasting are contained in the implementing regulation for European business statistics. A backcast series is therefore a published statistical product, with a method note, not a calculation that a reader can reproduce from the table alone. A reader should use the published backcast series where one exists, and should state that it is a backcast, with its method, when citing it.
Why are both methods needed?
Both methods are needed because the classification change breaks simple comparisons. A split breaks a series into parts whose shares are unknown from the table. A merge combines series whose separate trends are then hidden. A rename or renumbering breaks a series that looks continuous by title or digits. Double reporting shows the break directly, on the same period, under both versions. Backcasting removes the break from the historical series, so that a long trend can be read. Together, they let a user choose: compare the two versions side by side for the overlap period, or read a single backcast trend across the whole period. Neither method removes the need to state the version. A double reported figure is two figures, one per version. A backcast figure is a figure under the new version, produced by method, not an original observation under the new version.
What do the methods not decide?
The methods do not decide the code for any reader, and they do not decide tax, licensing, funding or collective agreement questions. They are statistical methods for published series. The entry for a specific unit is confirmed with the register that holds it. This site holds no register and decides no entry. A reader who treats a backcast figure as the original entry for a unit, or a double reported figure as two different entries, mistakes a statistical product for a register record. The register holds the entry. Statistics present aggregates. The two uses share a classification, and they must not be confused.
Double reporting shows the same period under both versions, backcasting re-presents history under the new version, and the published series with its method note is the comparable basis for reading across the change. ”