How did the division count change to 87?

The division count in ISIC Rev. 5, UN Statistics Division, is 87 divisions, reflecting divisions created, split, merged or retired in the move from Rev. 4. The count alone does not tell the reader which divisions changed, and it should never be quoted as if every division were new. For detail, the UNSD structure and the Rev. 4 to Rev. 5 correspondence table show the division level changes category by category.

Why does one total hide the real story?

Eighty-seven is a tidy number, and tidy numbers invite shortcuts. A reader might repeat it in a slide, a footnote, or a news paragraph, and leave the impression that Rev. 5 rebuilt the division level from scratch. The first paragraph of this guide warns against exactly that reading. Most divisions carried over in recognisable form, while a smaller set was reshaped. The total is the net result of several different kinds of change pulling in different directions, and a net result cannot show which way any single division moved. Anyone who needs more than a headline must leave the total behind and work through the changes one division at a time.

That discipline matters because divisions sit in the middle of the hierarchy. They are broad enough to anchor a published table, yet fine enough that a split or a merger changes what a time series means. A division total quoted without its history can therefore look stable while the ground underneath it has shifted. The careful move is to treat 87 as a starting point for checking, not as a conclusion.

What do created, split, merged and retired mean here?

Each of the four words in the opening paragraph describes a different fate. A created division is one that did not exist as a division in Rev. 4 and now stands on its own. A split division is one whose former scope was divided, so activities once counted together are now counted in more than one place. A merged division brings together scope that used to sit apart. A retired division disappears as a separate division, with its scope absorbed elsewhere or redistributed. These are not interchangeable labels for “changed”. A split breaks a series apart, a merger joins series together, and a retirement removes a heading a reader may still be looking for.

The practical consequence is that two divisions with identical titles in Rev. 4 and Rev. 5 cannot be assumed identical in scope. Nor can a division that vanished be treated as if the activities it covered vanished with it. The correspondence table is the tool that sorts this out, because it records the mapping division by division instead of summarising it in a single count.

How does a reader trace one division across the revision?

Start with the Rev. 5 division and read its full UNSD description, including the notes that say what belongs inside it and what is sent elsewhere. Then turn to the correspondence table and find the Rev. 4 rows that feed into it, or that it feeds into. Three questions keep this trace honest:

• Did this division gain scope, lose scope, or both?
• Does the correspondence row point to one Rev. 4 division, or to several?
• Does any part of the old division now sit in a different Rev. 5 division?

Answering those questions takes longer than copying the number 87, and that is the point. A short paragraph that names the direction of change for the divisions a table actually uses is worth far more than a correct total attached to an unchecked series. Where the correspondence shows a complicated mapping, say so in the methods note, and keep the table at a level where the comparison still holds.

Why should a division figure never travel alone?

Published tables are reused. A division figure copied into another report, chart, or dataset often loses its footnotes on the way, and the next reader has no way of knowing that a split or merger sits behind it. Keeping the version label, ISIC Rev. 5, UN Statistics Division, beside the figure helps, but the label alone does not explain a break. A line that records which divisions were affected, and how, is what lets a later reader judge whether two years can be placed side by side.

That habit also protects the writer. When a reader asks why a division total jumped or fell between revisions, the writer who traced the correspondence rows can answer from the record. The writer who only knows the total cannot, and must either guess or retreat to a vague statement about restructuring. Division level work rewards the slower path.

This guide explains a statistical classification and nothing more. It does not settle which code belongs to any particular business, what its principal activity is, or its registration, and for any specific record the office to ask is the national statistical institute or business register holding that record.