What does an old code in a new listing mean?

An old code in a new listing usually means that the listing has not been recoded, or that it was copied from a record made on the earlier version, and the code must be checked for its version before it is used. The publication date of the listing is not the version date of the code. The version is stated in the listing metadata, in the register record, or in the dataset behind the listing, and it must be found there, not assumed from the listing date.

How is an old code recognised?

An old code is recognised by its version, not by its appearance. A NACE Rev. 2 code and a NACE Rev. 2.1 code can look identical as strings, where the category did not move, and they can look different where the category split, merged, or moved. A listing that states its version can be recognised at once. A listing that does not state its version must be checked against the register, using the identifier for the unit, and the register record will state the version in use. A listing that shows data for an earlier reference period, or that was compiled from an earlier register extract, will show old codes for that reason, and the codes are correct for the period and the extract, even though they are old.

What are the three causes of an old code in a new listing?

Three causes cover most cases, and they should be distinguished, because each calls for a different next step. First, the listing has not been recoded: the register moved to the new version, but the listing still shows the old extract. The next step is to open the register and record the new code. Second, the listing was copied from an old record: the listing shows the code as held at the copy date, and the copy date is earlier than the recoding. The next step is the same, with the copy date recorded. Third, the code is genuinely on the old version because the record it describes belongs to the old period: a historical dataset, a report for an earlier reference period, or a register extract made before the change. The next step is to cite the code on its own version, with the reference period stated, and not to recode it by assumption. The table below shows the three causes and the next step for each, so that a reader does not treat a historical code as an error, or an unrecoded listing as a historical record.

Cause	Evidence	Next step
Listing not recoded	Register shows a new code, listing shows an old code	Record the register code, with version and date
Copied from an old record	Listing copy date precedes the recoding	Record the register code, note the listing copy date
Record belongs to the old period	Reference period is on the old version	Cite the code on its own version, with the period stated

The table is a diagnosis, not a correction. A listing that is diagnosed as unrecoded should be quoted, where it is quoted at all, as an unrecoded listing, with the register code given alongside it. A listing that is diagnosed as historical should be quoted on its own version, and should not be updated by the reader.

How is the correspondence table used for an old code?

The correspondence table is used to trace an old code to its new place, and to record the mapping type. Eurostat states on its correspondence tables page that tables between different versions of the same classification describe the detailed changes that have taken place in the revision process, and on its transition page that the table between NACE Rev. 2 and NACE Rev. 2.1 describes, for each NACE Rev. 2 heading, how activities currently classified therein should be classified in NACE Rev. 2.1. Where the table shows a one to one mapping, the old code traces to one new code, at the level shown. Where it shows a one to many mapping, the old code traces to more than one new code, and the table alone does not pick the new code for a specific unit. Where it shows a split, the explanatory notes for the new classes must be read, and the register entry for the unit must be confirmed, before any new code is stated for that unit. The table traces the category. The register holds the entry.

What does an old code not prove?

An old code does not prove that the business is out of date, that it has stopped trading, or that it has failed to comply with a change. It proves that the record consulted shows a code on an earlier version, for one of the causes above. The Central Statistics Office states that its Business Register completed updating the coding of enterprises to NACE Rev. 2.1 during 2025 (FIXED, as stated by the Central Statistics Office), and that publication moves gradually from January 2026 onwards (FIXED, as stated by the Central Statistics Office). During a gradual change, old and new codes coexist in different records, and an old code in one record is expected, not an error. A reader who treats an old code as a fault, without checking the version and the register, will misjudge both the listing and the business, and the judgement should be withdrawn once the version is checked.

Where is an old code checked?

An old code is checked in the register that holds the record, using the identifier for the unit, and in the correspondence table, for the mapping type. The meaning of the old category is checked in the explanatory notes for the old version, and the meaning of the new category in the notes for the new version. The entry for a specific business is confirmed in the register, on the version the register currently uses. A code that cannot be traced in the table, because the table version is not stated or the category is not listed, should be recorded as untraced, with the date of the check, rather than being traced by assumption.

Record the version from the source, trace the category in the correspondence table, and confirm the entry in the register that holds it.