How do national implementations follow Rev. 5?

National implementations follow ISIC Rev. 5, UN Statistics Division, by deriving a national classification from the international structure, usually keeping the ISIC categories and adding national detail where local statistics need it. Each country sets its own timetable, so Rev. 5 may appear in one country’s releases years before it appears in another’s. A reader must therefore check the national implementation date for the dataset in hand, not assume the March 2023 endorsement date means immediate national use.

Why do national timetables differ so much?

The opening paragraph gives the reason in one sentence: each country sets its own timetable. Endorsement of an international standard is a starting signal, not a switch that flips every statistical system at once. A national office must adapt questionnaires, retrain coders, update registers, revise back series or document breaks in them, and align the classifications used alongside ISIC, such as those for products and occupations. That programme of work takes different lengths of time in different systems, and it competes with the ordinary business of publishing statistics on schedule. Two countries can therefore be years apart in Rev. 5 use while both fully intending to adopt it.

A reader who expects uniformity will misread the landscape. A dataset labelled with a recent year may still rest on an older national classification, while an older dataset from another country may already use Rev. 5. The calendar year of a release says little on its own. The classification label, and the national implementation date behind it, say a great deal.

What does deriving a national classification involve?

Deriving a national classification from ISIC Rev. 5, UN Statistics Division, usually means keeping the international categories intact and adding finer national detail where local statistics need it. The international structure provides the shared skeleton: the same sections, divisions, groups and classes, with the same definitions, so that results can be compared across countries. National detail then sits below or alongside that skeleton, splitting a category that matters locally into smaller pieces, without disturbing the international totals that those pieces add up to.

That design is what makes comparability possible without forcing every country to publish at the same level of detail. A country with a large local interest in a particular field can count it finely for its own purposes, while still reporting the international category that other countries report. Problems arise only when national detail is mistaken for international structure, or when a national category is quoted as if it were an ISIC category. Keeping the two layers visibly separate, in documentation and in table headings, prevents that confusion.

Where should a reader look for the national date?

The national implementation date is not always prominent, which is why the opening paragraph tells the reader to check it rather than assume it. Useful places include the methods note attached to a release, the classification documentation published by the national office, and the notes that accompany a business register or a survey dataset. A release may also state the classification version in a table footnote, or in a separate methods document that the release links to. If none of those places names a version, the honest conclusion is that the version is unknown, not that it must be Rev. 5 because Rev. 5 is current internationally.

This check becomes especially important around transition years. A single publication may contain some tables on the old basis and some on the new, or a series may switch basis partway through. In those cases the implementation date is not one date for the whole release, but a date per table or per series, and the reader needs to record it at that finer grain. A methods note that records the switch point saves every later reader from reconstructing it.

What happens while a country is mid-transition?

Mid-transition periods produce mixed datasets, and mixed datasets demand explicit handling. A business register may be recoded in stages, a survey may move to the new classification before the national accounts do, and different products from the same office may sit on different bases for a time. None of this indicates failure. It indicates a large system changing its reference structure while continuing to publish, which is a difficult operation carried out in public.

For the reader, the practical rule is to treat every table as carrying its own classification basis until shown otherwise. Do not carry a Rev. 5 label from one table to the next table in the same publication without checking. Do not assume that because a country announced Rev. 5 adoption, every dataset it publishes has already moved. Announcements describe intentions and programmes. Tables describe what was actually used for that release, and tables are the evidence that counts.

This guide stays at the level of general statistical description. It does not fix a code, a principal activity, or a registration outcome for anyone reading it; those questions belong to the national statistical institute or business register responsible for the record in question.