Recoding of the Enterprises in the Business Register


The transition to a new classification is normally a very complex process as it involves virtually all aspects of the compilation and presentation of the relevant statistics. This is even more true of the transition to a new classification of economic sectors, since classifications of this kind are used in a wide variety of statistics and a high level of coordination is therefore required. An important basic prerequisite for the transition to a new classification of economic activities is the recoding of all statistical units in the business register to comply with the new classification. Furthermore, the units have to be coded according to both the old and new classifications in parallel for a certain period of time. The transition process obviously also needs to include all those institutions that list the units in their registers according to the ÖNACE classification. This applies in particular to the tax administration and social insurance registers since these registers provide the initial assignment of new units according to the ÖNACE classification in the process, for example, of maintaining the statistical business register. These institutions also use the ÖNACE classification for their own purposes and publish statistical data according to ÖNACE.

On the basis of the European guidelines for the implementation of NACE, we will first introduce the concept of the transition to ÖNACE 2008 in the business register and, secondly, present considerations as to how Statistics Austria can support the transition to ÖNACE 2008 and the ongoing implementation of ÖNACE 2008 in the various administrative registers.

Concept of transition in Statistics Austria’s business register

The total amount of units to be recoded consisted of the sum of enterprises listed in the business register and their establishments and local units, with each unit being assigned a six-digit figure (subclass) under ÖNACE 2003 according to their economic focus. Secondary activities were also recorded in many cases, also at six-digit level. To simplify matters, however, reference will only be made in the following paragraphs to the level of enterprises, which numbered just under 380 000 by the end of 2006. Here too, only the basic concepts of transition will be addressed.

The analysis of relationships between current and new classifications formed the starting point for considerations of recoding. Since this was a full-scale revision, it could be expected that the two classifications would differ significantly in structure and that simple, virtually automatic recoding would not therefore be possible in many cases. In addition, it had to be ensured that the recoding process does not have a negative impact on the quality of ÖNACE coding.

In principle, the relationship between elements in two classifications may be one of the following four types:

  1. 1:1 cases: One element in the old classification can be found in the new classification in the same way.
  2. 1:n cases: One element in the old classification has been subdivided into two or more elements in the new classification.
  3. n:1 cases: Several elements in the old classification have been aggregated to form one element in the new classification.
  4. m:n cases: Parts of different elements of the old classification have been reassembled so that the new elements are not simply aggregates of old elements or disaggregates of one old element. These are typical configurations in classification revisions.

It was therefore clear that only 1:1 cases and n:1 cases could be recoded automatically. Specific additional information was therefore required or certain assumptions needed to be made for all the other configurations.

Based on these concepts the enterprises of the business register were recoded as follows:

  1. Around 50% of enterprises were recoded automatically as they were assigned to classes that had 1:1 or n:1 cases.
  2. A further 17% or so of enterprises were recoded automatically as well because the existing subclass assignment clearly indicated the new class (subclass) of ÖNACE 2008 to which they should be transferred.
  3. Just under 2% of the remaining enterprises were recoded automatically too, because they were subject to PRODCOM reporting requirements, thus providing information at product level about which activities they were engaged in and which was the main activity.
  4. Just under 3% of the remaining enterprises were also recoded automatically although they are assigned to elements that have m:n relationships. It is highly probable, however, that the activities responsible for the complex m:n relationship structure are not relevant in Austria. A 1:1 relationship is therefore assumed in these cases.

In total, therefore, just under 73% of enterprises were reclassified automatically on the basis of the relationships between the old and new classifications or on the basis of existing information alone. The remaining 27% of enterprises were divided into two groups:

  1. For around 60% of these enterprises, it was assumed that the information on the enterprise’s activity required for the recoding could be determined by internal research. Research sources included the Internet, the Yellow Pages, trade directories and similar tools as well as expert knowledge.
  2. All the remaining enterprises – around 42 000 – had to be contacted, however, and asked to provide information on their activities so that they could be correctly classified according to the new classification.

The enterprises either received a written questionnaire or, in relatively straightforward cases, were contacted by telephone. They were asked about the various activities in which they were engaged and the proportion of turnover accounted for by each activity in order to deduce the main activity of the enterprise. The questionnaires were tailored to the specific classification area, with probable activities already being listed. The survey was conducted during the first three quarters of 2007 and took § 25/4 of the Federal Statistics Act 2000 as its legal basis.

Transfer of the new assignments to the administrative register

The recoding of the enterprises in the business register was finalized on time in the autumn 2007. As a result, it was possible to analyse the new data available and drew the new random samples as well as to transfer the new classification assignments for each unit to the administrative registers (tax, social insurance, etc.) to save these administrative offices the time and effort of recoding the units.

This also raised the subject of a further requirement of the Federal Statistics Act, namely that Statistics Austria’s classificatory assignments are to be used in all areas (§ 21 Federal Statistics Act 2000). It is not enough, of course, merely to adopt the new classificatory assignments for the total set of data; there must also be ongoing adjustment for new or amended cases. Since the assignments currently diverge to a certain extent, there would be a break in the relevant statistics if Statistics Austria’s classificatory assignments were used. As the transition to the new classification creates a break in the time series in any case, this break can quite easily include the break caused by adoption of Statistics Austria’s classificatory assignments.

For the classification of new units due to take place from 2008 or in the event of changes of the main activity under ÖNACE 2008, Statistics Austria will provide the required classification aids in electronic form and is also willing to hold information or training events.