Session 6 - Quality of the Business Register, Chaired by Sven Egmose (DK)

Statistics Denmark27 January 2000

Business StructureSEG/-

Akt.nr. 3590

1

13th Roundtable.

Session 6 - Quality of the Business Register, chaired by Sven Egmose (DK).

The session featured 7 papers.

The papers addressed 4 main themes within quality

- quality problems regarding the sources for the BR

- overcoverage because of falsely active units

- measurements of the quality of the BR

- means to improve the quality of the BR .

Very brief abstracts of the papers and the following discussions are given below.

1. Germany

The paper describes the problems of measuring the quality of the German Business Register. Because of a very restrictive legislation it is impossible to make surveys to check the quality and all the relevant administrative registers have to be used as input to the Business Register.

Discussions points:

It was asked whether the administrative registers do divide the population between each other and if a co-operation exists between the registers, and the answer was that there is no collaboration between the registers.

To a question on the use of standardised variables in the different registers the answer was that the Federal Statistical Office has no influence on the content of the registers.

2. France

The paper calls attention to the fact that some units are considered to be active while they are not. This is a problem for all BR's. In the paper an estimation is made of the dimension of the problem in the French BR, and the falsely active units are analysed according to size, activity code etc.

Discussion points:

It was asked whether there is correlation between the variables used for analysing the falsely active units. There might be some correlation. Nevertheless, at least size of the units is a significant variable explaining the occurrence of falsely active units.

Against a proposal to use VAT data as a measure to remove the falsely active units it was argued that VAT data are of poor quality for the small units.

To a question on the existence of data on more detailed activity code level it was answered that the analyses was not made on a more detailed level than described in the paper.

3. New Zealand

The paper describes three quality studies regarding the BR carried out by Statistics New Zealand. The main results were that quality was good regarding coverage (of places of work), activity codes, and data on lifecycle codes. The paper argues that the quality standards must be fixed in consultation with the users.

Discussion points:

It was agreed that the information from respondents in some cases are rather inexact. To a question on whether the respondents were contacted in the process of recoding activity code etc. it was answered that it was only done very seldom.

It was also told that no measurements were made of the quality of the work of the coding personnel.

It was stated that the proportion of false activity codes was very low. To a question about the reason for this it was argued that there might be a connection to the fact that only large enterprises were studied.

4. Denmark

The paper suggests some main quality features for the BR. The paper also discusses different ways to measure the quality, and describes the main results of a quality measurement of the Danish BR. The actions taken to improve the quality of the Danish BR are briefly gone through.

Discussion points:

It was argued that quality and costs must be weighed, and it was agreed that cost benefit considerations are very relevant to present to users when discussing quality with them.

In the future the quality of the statistical BR to a very high degree depends on the quality of the new administrative BR in Denmark. To a question on quality standards for the new administrative BR in Denmark it was answered that the quality standards for this register was not fixed yet.

5. Australia

The paper describes the co-operation between the ABS and the Taxation Office. The collaboration with the Taxation Office in order to improve the quality of the activity coding is very important. The results of the extended co-operation is highly heartening as studies have shown a considerable improvement of the quality of the activity codes in the BR.

Discussion points:

The studies of the quality of the activity code was made on ANZSIC class level. Questions were asked to more detailed activity code level but the answer was that the studies were not made on a more detailed level.

The Australian BR rely - like most other BR's - heavily on administrative data. Attention was called to the fact that this means that administrative changes may affect the continuity of the statistics. Hence the Statistical Offices have a big interest in following and having influence on the administrative concepts used.

6. Sweden

The paper describes the quality aspects of the work with the new Swedish BR. By the term quality is meant both the quality of the product and of the process, and plans are to measure both dimensions of quality in the future. Sweden is following the Total Quality Approach which among other things means that the starting point for content and quality is user needs and expectations.

Discussion points:

It was questioned who is going to run the metadatabase of the BR. The answer was that this should be done in close co-operation with the institutions running the source registers.

It was asked how to keep a rein on the cost of running the BR. The answer was that it will be a part of the process to find out where you get 'most for least' and effort will be devoted to such fields.

Regarding the unit types in the register it was stated that besides other unit types also the enterprise group will be a unit type in the new Swedish BR.

7. The Netherlands

An important political goal is to minimize the burden on enterprises, and in order to achieve this the Dutch government has launched a policy of creating a system of 'authentic registrations'. The paper refers to this and describes one of the means to achieve the authentic registration, viz. to build up a single BR to be used for both administrative and statistical purposes.

Discussion points:

In the paper the wording 'authentic register' is used and it was asked who is to decide when the BR has this status. The answer was that the register will not be labelled as such as an authentic register. The meaning was that when the register is used for all purposes it is an authentic register.

Regarding the access to the authentic BR register it was stated that there should be public access to the register but only for those who had a need to know some features in the register.

1