How were new activities accommodated?

New activities in ISIC Rev. 5, UN Statistics Division, were accommodated by creating new categories where the activity field was large and distinct enough, by revising definitions where an existing category could stretch, and by using residual categories where detail was not yet justified. The 18 new intermediation classes are the clearest example of the first route. Accommodation is a judgement about statistical need, not a recognition award: an activity gets a category when counting it separately improves the statistics.

What question decides whether a new category appears?

The opening paragraph frames accommodation as a judgement about statistical need, and that framing is the key to the whole guide. The question is never whether an activity is interesting, novel, or widely discussed. The question is whether counting it separately improves the statistics enough to justify a new category. Size matters, distinctness matters, and durability matters. An activity field that is large, clearly separable from neighbouring fields, and likely to persist earns its own heading. A field that is small, fuzzy at the edges, or likely to fade is better handled inside an existing category or a residual one, until the evidence grows stronger.

This test protects the classification from two opposite failures. Too eager, and the structure fills with headings that few tables can populate reliably, producing thin, noisy series. Too reluctant, and genuinely large new fields disappear inside old headings, where no reader can find them. Rev. 5 tries to steer between those failures, and the three routes in the opening paragraph are the steering mechanism.

Why do new categories suit the intermediation classes?

The 18 new intermediation classes are the clearest example of the first route because intermediation grew into exactly the kind of field the test describes: large, distinct, and persistent. Intermediation sits between parties rather than producing a good or delivering a final service in the traditional sense, and that in-between position is what older categories struggled to place cleanly. Giving such activities their own classes lets tables show their scale directly, instead of burying them inside headings whose titles suggest something else.

The number 18 also matters as a signal of granularity. It shows that the revision did not settle for a single token heading for intermediation, nor did it fragment the field beyond what compilers could populate. Eighteen classes suggest a field examined in detail, with distinctions drawn where the data could support them and withheld where they could not. A reader who sees those classes in a table should therefore read them as a deliberate measurement decision, not as decoration.

When is revising a definition better than creating a category?

The second route, revising definitions where an existing category could stretch, suits activities that are new in form but continuous in substance with something already counted. A definition can be widened, its explanatory notes updated, and its includes and excludes lists adjusted, so the new form finds a home without disturbing the structure around it. This route preserves series continuity, which is a real benefit: a revised definition keeps a time series alive, while a new category breaks it, at least at first.

Stretching has limits, and recognising them is part of the judgement. A definition stretched too far becomes vague, and a vague definition collects unlike activities under one title until the heading means less and less. The test, again, is statistical: does the stretched definition still describe a coherent field that compilers can code consistently and readers can interpret? If the answer is yes, revision is the economical choice. If the answer is no, a new category, or a residual placement, is more honest.

What work do residual categories do?

Residual categories are the third route, and they are often misunderstood as dumping grounds. In a statistical classification they have a precise function: they hold activities that are real but not yet large or distinct enough for their own heading, so that totals remain complete without pretending to a detail the data cannot support. A residual category says, openly, that a field exists here and is not yet separately measured. That statement is far more useful than forcing the field into a neighbouring heading, where it would distort that heading’s meaning, or omitting it, which would break the totals.

Residuals also act as watch posts. If a residual category grows steadily across releases, that growth is evidence that a future revision may need to split it. If it stays small, the decision to withhold a separate heading is vindicated. Either way, the residual has done statistical work: it kept the accounts complete and it generated the evidence for the next judgement. Readers who skip residuals when reading a table miss this early signal of structural change.

This guide describes how a statistical system organises new fields. It offers no determination of any reader’s code, principal activity, or registration, and such a determination should be sought from the national statistical institute or business register holding the relevant record.