What happens when a level is not split?

When a level is not split, the hierarchy is preserved by repeating the heading at the lower levels, with codes formed by appending zeros. A section that is not split has one division, one group and one class, each with a heading identical to the section. A division that is not split has a group and a class with the same heading as the division. The repeated heading is the rule for an unsplit level, not a duplication error in the table.

Which unsplit cases does the manual name?

The manual names three patterns. First, if a section is not split at all, there is one subordinated division, one subordinated group and one subordinated class, each with a heading identical to that of the section, and the sole current case is section V, for activities of extraterritorial organisations and bodies. Second, if a division is not split at all, the group and class must have the same heading as the division, with codes composed of the division code appended by the digit zero at group level and the digits zero zero at class level. The example given is division 75 for veterinary activities, group 75.0 and class 75.00, all with the same heading.

Third, if a division consists of a single group which is split at class level, the group must have the same heading as the division, with the group code formed by appending a zero to the division code. At class level, none of the headings may be the same as the group heading, because the classes divide the field into detailed content that must be reflected in each class heading. The example given is division 11 for manufacture of beverages, group 11.0 with the same heading, and classes such as 11.01 for distilling, rectifying and blending of spirits, 11.02 for manufacture of wine from grape, and 11.03 for manufacture of cider and other fermented fruit beverages.

Pattern	Code example from the manual	Heading pattern
Section not split	Section V	Division, group and class share the section heading
Division not split	75, 75.0, 75.00	Group and class share the division heading
Single group, split at class level	11, 11.0, 11.01 to 11.03	Group shares the division heading, classes have distinct headings

Why does the classification keep unsplit levels?

The classification keeps unsplit levels because it is balanced. The manual states that NACE is a balanced classification, even where positions are not subdivided at all. Balance means that every position has a place at every level below it, so that statistics can be aggregated and disaggregated consistently. A table at class level can be aggregated to group, division and section level without a special rule for an unsplit branch, because the unsplit branch already has a category at each level. A table at section level can be disaggregated to class level in the same way, because the class exists even where it repeats the section heading.

Balance also preserves comparability over time. If a section is unsplit in one version and split in a later version, the structure table for the later version will show the new divisions, groups and classes inside the section, and the correspondence table will show how the earlier single category relates to the new categories. If the unsplit level had been omitted, there would be no category at the lower levels to correspond to the new categories, and the comparison would require an ad hoc rule. The zero code pattern avoids that problem by giving the unsplit branch a code at every level from the start.

How are unsplit codes read in a dataset?

An unsplit code is read in the same way as any other code: split at the dot, find the division, find the section, and read the note. The zeros in the code are part of the code, and they must not be dropped. Code 75.00 is a class code, not a division code written with extra zeros, and code 75.0 is a group code, not a division code. Dropping the zeros would change the level, and aggregating a dataset by the shortened code would mix levels. A dataset that shows 75 in one row and 75.00 in another is showing the same branch at two levels, and the two rows must not be added together as if they were separate activities.

The note for an unsplit category is also read in the same way as any other note. The manual states that the content and scope of each position is defined through detailed explanatory notes. An unsplit class has a note, and that note states what the class includes and excludes, even where the heading repeats the division heading. A reader who assumes that a repeated heading means there is no note, or no distinction to be made, would miss the definition. The note must be read, and the index entries for the class must be checked where they exist.

What does an unsplit level not mean?

An unsplit level does not mean that the field has only one activity, and it does not mean that the field is unimportant. It means that the classification, as adopted, makes no finer distinction inside the field at that level. The field may contain many activities, all grouped in one class, because the criteria for a separate class, such as prevalence in most member states or particular importance in the world economy, were not met for a finer split, or because a finer split was not explicitly required by new concepts. The manual states that efforts were made to maintain the structure in areas that did not explicitly require change, and that overall characteristics were not changed through Update 1. An unsplit branch is therefore a deliberate part of the adopted structure, not a gap in the drafting.

Check the code for appended zeros, read the note for the category at the level shown, and confirm any entry with the national statistical institute or the business register that holds the record.