导航|地址|刷新
Summary and Analyses of the Responses to…























Chinese(GB)

RSS

Organizational Chart









Data

Communiqués

Legal Framework

Classifications & Methods

Latest Releases

International Cooperation

Understanding Statistics

Agency





Home > papers


Summary and Analyses of the Responses to the Questionnaire

SRC-3842005-01-13 09:37Print| Large| Medium| Small



Summary and Analyses of the Responses to the Questionnaire

Hugues Picard,INSEE, France
Jean Ritzen, Statistics Netherlands?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />

How does the Business Register coordinate Economic Statistics or does it contribute to this? What are the advantages and the drawbacks of the various registers and of the various organizational positions of these in the different countries? What are the plans for future organizational developments and for future role and content of the Business Registers? These questions were at the basis of a questionnaire, the answers to which might induce some discussions during the Roundtable meeting and result in important issues for which more in-depth considerations could be made in future discussions and meetings.

The questionnaire was sent out to all known possible participants. Fifteen countries have responded (including six countries which will not be represented in the Beijing meeting). We are thankful to the respondents for the quality and the interests of their answers and remarks.

Introduction

A « Statistical Business Register » (SBR) is to serve the statistical processes as a part of them, which implies the work performed by the agents in charge of the SBR. Another important feature that influences their tasks is linked with the availability of information for the creation and the maintenance of the SBR. The SBR is not a file or list in its own, it should be seen as this file in combination with the belonging methodology to meet the goals. It is better to speak of a business register system. And this system is part of the total system of economic statistics.

What do statisticians expect from the SBR? An up-to-date list of statistical units, which is exhaustive, without double counting or falsely active units, including well-defined rules and procedures for maintenance and processing. This is the basic requirement. The features imply:

¨the creation of the statistical units the statisticians want to deal with;
¨the selection of a list or of statistical units according to some criteria of size and/or activity and/or location. These characteristics must be added to the statistical units for this reason;
¨inclusion of information on where to collect the relevant information for a selected or sampled statistical unit;
¨to be a benchmark for making the necessary adjustments or corrections when usingadministrative files in the production of statistics.

Besides these purely statistical purposes, some countries use the SBR also as a tool to manage and measure the statistical burden they impose upon enterprises. Here the SBR is also a tool for administrative management of data collection processes.

The main benefit of the ideal position and use of the SBR lies in the resulting quality of statistics. Good use results in full comparability of statistical output from different surveys regarding the same reference period, and good comparability of similar surveys over time (time series). Here we see the co-ordinating role of the SBR. This requires well-developed statistical methodology, and also discipline of statisticians who consider the quality of the whole package of statistical output above quality of the individual statistic they are responsible for.

The main challenge for the BSR agents is to transform into statistical information the data collected from surveys or from administrative declarations by legal units (natural persons or companies or parts of these) which declarations are legally based on accounting or social concepts. Sources of information are necessary for the creation and maintenance of an SBR These can differ for the different countries in availability, costs and effectiveness.

In most countries there is a special link between the SBR and the compilation of economic demographic statistics. Although in a perfect situation economic demographic statistics may be tabulated from the SBR directly, this is usually not the case because of SBR imperfections.

SBR imperfections also have repercussions for regular statistics based on the SBR. These can regard the units themselves and the characteristics assigned to the units. How to manage these imperfections is to be seen as an important issue, to be explored further. The survey was not intended to hold a survey on quality of the statistical business register or to find criteria for this. Quality aspects are dealt with in another session of the Roundtable meeting, although it is neither possible nor desirable to disregard quality aspects.

For international comparability of statistical output we need basic international co-ordination. Especially basic conditions need sufficient attention, as people realise when it comes to statistical business registers. Statistical business registers are on the agenda of e.g. the UN and Eurostat[1] . The annual Roundtable should be seen as a tool for discussion to reach further convergence on principles and to discuss new developments that are important for the SBR.

The sources for creating and updating the SBR

After reading the answers to the questionnaire, countries can be divided into three groups according to the data they have available for the construction and maintenance of the SBR:

1.the countries, which have no or very limited access to administrative files;
2.the countries which have access to administrative files and derive their SBR from these files. This group can be subdivided into two categories depending on whether one file has a pre-eminent role (usually a Tax Collecting Administration file) or more than one administrative file has an important role in the making of the SBR;
3.the countries, which have an official Business Register from which the SBR is derived. In this we see two categories again: countries where this register is kept in the statistical agency and countries in which this exists outside the statistical agency.

In general, the countries of the first group create their SBR from a (regular or not) Census of enterprises. The Census is organized so that the statistical units used in the country can be easily derived, and some questions to the respondents are used for creating the variables needed in the SBR, including the main activity code (and the principal secondary activity codes). The SBR is updated from the various annual surveys as well as from any information the SBR agents may collect from newspapers, telephone books, professional information or professional lists, etc.

The answers show the countries using administrative files do so to create and update the SBR. However, some of them use only one administrative file, generally the file of the Tax Collecting Administration, while others use several reference files, usually the Tax Collecting Administration file and one or several files from Social Security and National Health Insurance Administrations.

The agencies where an official administrative Business Register exists use this database as a reference, but they may also take some supplementary information available in other administrative files.

Generally, the information included in the reference administrative file or in the reference BR, even supplemented by other administrative files, is not sufficient to obtain all the information required in the SBR. This is particularly the case with regard to the construction of statistical units. Several solutions are applied:

¨contact the company and ask for the relevant information, which sometimes includes a request for an estimation of the value of consolidated data for the statistical unit (for example, turnover may not be totalized by adding for units in a cluster of linked companies within enterprise groups);
¨try to find information from various sources (existing surveys, newspaper, company reports,….) or make some imputation;
¨try to have one of the administrative data collectors (one of those the file of which is used for the SBR) include some special questions for the SBR in the regular administrative form that the companies have to fill in.

All these procedures have their drawbacks or may not be possible.

The statistical offices are reluctant to contact the companies, since this increases the statistical burden. Generally, only the major companies are contacted, usually within the framework of the profiling operations. But this procedure is not widespread for the time being: only half of the countries that answered the questionnaire declare making profiling; furthermore, some make it on a very narrow scale, but the great majority declare that developing profiling is in their plans for the future. For smaller companies, which are not profiled, some countries have designed specific rules (most of the time automatic) for constructing statistical units from the information they obtain on a regular basis from regularly used sources.

The second procedure is not very often cited. It is true that it cannot lead to an established procedure since the information arrives at random in the source consulted. However, the research of information in every place can lead to the improvement of the SBR, although unsystematically. In countries where the statistical agency has no access to administrative data and very limited resources for the SBR and for carrying out regular enterprise surveys, this is at least a valuable alternative for improving quality.

As to the third procedure, only a few statistical offices declare having some influence on the data collected by other administrative agencies for the establishment or maintenance of the BR. Some seek to obtain a good main activity code; in that case, close relations are established with the codifying administrative agency, with possibility of exchange of staff or design of common codification software. In one case, the tax agency is mentioned as the agency responsible for collecting information for statistical purpose only, thus for the Statistical Office. At last, one respondent indicates that the Statistical Office is involved in the definition of the legal units and in the delineation of local units, and participates also in the specification of some quantitative data collected in administrative forms, such as the number of employees.

The identification numbers

The registration of the identification numbers in the administrative files of the components of the statistical units is important in the process of creating and up-dating the SBR. The knowledge of these numbers eases the use of administrative files since the administrative unit(s) that have to be considered for obtaining relevant data for a given statistical unit are known. This is even more efficient when a unique identification number is used throughout the various administrative files used by the statisticians.

Half of the respondents to the questionnaire indicate that a unique identification number is used in most or all administrative files, and, in the majority of countries, this number is the one issued by the Tax Administration mostly. One country indicates that there is no unique identification number. Furthermore the question is not relevant for statistical offices that do not have access to administrative data. The issue not raised in the questionnaire, refers to the obligation by administrations to use a unique national identification number in the administrative files, and if so, the issue of the quality of the maintenance of the identification number by the various administrations.

The questionnaire also includes a question about the dissemination of the identification number of the SBR units. The answers indicate that this question is irrelevant, even when the statistical office issues the unique administrative identification number. For management reasons, a specific number identifies the statistical units, which may be the identifier of the administrative file or the administrative BR used to create and update the SBR only in the case of a one to one transposition from legal unit into statistical unit. Statistical units are units constructed for statistical purposes. Even if the SBR is public, these units only become important for companies if they also can obtaininformation on the administrative units as composing parts of these (legal units within enterprise groups, for example); some companies know also the statistical units resulting from the profiling of the group or sub-group to which they belong. So there is only one reason to provide the identification number in the SBR: this is the case when an administration agrees to include some questions on information for a statistical unit defined by the Statistical Office in its regular forms filled by businesses. But this all seems very theoretical and in many cases we are confronted with restrictions from confidentiality policy.

In some countries, a special administrative BR is established in the Statistical Office as an intermediate product between the administrative file(s) or the administrative official BR and the SBR. This is the basis for the construction and maintenance of the SBR. If the legal and other administrative units, which are included in this intermediate BR, are related to the SBR with a specific identification number, then there is no reason for the Statistical office to promote this number for use in administrative files. This remark, along with the previous explanations, clarifies why most of the responses to the question on the use of the SBR identification number in administrative files are “never” or “very rarely”.

What about the uses of the SBR in compiling statistics?

Most of the respondents declare that the SBR is used to follow a unit over time. Nearly all countries have defined continuity criteria for the units managed in their SBR. The distinction in the use of continuity rules for annual statistics and for sub annual statistics has not been indicated; this is an important issue however, since the procedures for dealing with continuity may not be the same, in particular because in the case of sub-annual series, which are to be obtained rather rapidly, all information for applying the continuity criteria may not be available.

The use of the SBR for amending data when processing administrative files is rather widespread or will be soon introduced. Only very few respondents have no plans in this direction (e.g. because of no access to administrative files). Three main issues in processing administrative files can be cited:

¨transformation of administrative data into statistical data; this is generally an application of the SBR reported by all users of data in administrative files for compiling statistics along the SBR;
¨control of exhaustiveness of the administrative file (and imputation for missing units); from the comments, it seems that this is a use of the SBR in statistical processing for almost all agencies who make use of data in administrative registers for compiling statistics. It is an exception that agencies do not make any control of exhaustiveness (in terms of missing units);
¨inclusion of supplementary information (not present in the administrative file but in the SBR) in the processing database for compiling statistics; some countries explicitly report the enrichment of the database with data included in the SBR (for example, addition of the number of persons employed to the data on turnover and profit derived from an administrative file). It may be concluded that this application is generally practised.

Most of the respondents declare the direct use the SBR for economic demographic analyses. In fact, the comments show that the “direct” use of the SBR depends on what is included in the SBR. Furthermore, in order to obtain some demographic statistics, it may be necessary to gather information from other sources, besides the SBR (turnover, sub-classification of the official nomenclature distinguishing, for example, the various types of commerce,…). In fact, some analyses may require a (special) sample survey, in particular if good statistics on death have to be obtained. It can be concluded that in countries where demographic statistics are compiled the SBR plays a major role for the purpose, but not the only one.

Lists and samples

As expected, at all agencies the SBR is the usual base for making lists of units according to specified groupings and for deriving sample frames for business surveys. Most agencies designed compulsory sampling procedures and part of these is also the use of the SBR as the sampling frame. The comments show that in many places, the samples are drawn by a specific unit (may be included in the SBR unit), which permits to coordinate (make consistent and integrate) the samples of the various economic surveys and ensures the control of the statistical burden on enterprises (few mentions). One country (Canada) mentions having a semi-centralized organization:

“BR creates a Statistical Universe File (SUF) for each business survey. It is based on the survey population characteristics (e.g. NAICS). Surveys select their sample from their SUF and return it to BR. BR creates a Survey Control File (SCF) for the survey to keep track of the units in the SUF, which have been sampled in the past, and those sampled at the moment. Surveys maintain contact information in BR. BR prepares a Statistical Interface File for collection arrangements. Surveys send BR a file of all records they actually contacted. BR uses this information to keep track of response burden and non-response. Surveys send survey feedback and, when data capture is finished, a file with response codes.”

Feedbacks from surveys

As to the survey feedback, it seems that information, which may lead to modifications in the SBR, is usually returned from the business surveys to the SBR. However, if this practice appears to be a rule, in many statistical agencies it may be limited to a small number of surveys, or left to the attention and willingness of the persons in charge of a given survey. It should be noted that the use of the feedback information is generally not transferred in automatic procedures: the eventual change in the SBR is firstly controlled and checked by the SBR agents.

Activity code

As to the activity code, there are various practices in assigning this code to the units and the corresponding responsibilities. At the creation of the administrative unit, the assignment of the activity code is initially made by the statistical agency in four of the thirteen responding statistical offices. For the other nine countries, the code is assigned by one or several non-statistical administrations. In some cases, there are official relations between these non-statistical agencies and the statistical office, in order to co-ordinate the decisions taken, because, in all countries, the statistical office can overrule the decision of the non-statistical agency. Of course, the statistical office is also responsible for the assignment of the activity code at the creation of the statistical units, which do not have a one-to-one correspondence with an administrative unit. Ultimately the statistical office itself is responsible for the quality of statistics and the quality of the assigned activity code as used in statistics. In practise a big dilemma is inherent in this statement. Basically groupings in classes are defined from the assignment of the activity code.

The initial codes are eventually modified, if (checking or verifying) information from various sources (surveys, newspapers, company reports,…) becomes available. For many countries, the modification of the activity code in the SBR is done continuously in processing procedures including checking the information against already available information and plausibility checks; one country modifies the SBR once a month while four modify or adapt only once a year.

There are also differences in the production of sampling frames. Some countries preparea “frozen” version of the SBR once a year, which will be used to make the samples for the business surveys of that year, while other countries use the SBR as it is at the sampling date. In any case, the data on the date of the change of activity and the date of the introduction of the change in the SBR are included in most SBRs, which permits reconcililiation of the two procedures. The introduction of time stamping changes in the SBR gave a new dimension to the SBR for the use as a co-ordinated reference and sampling frame.

The question was raised on the consistency between the sampling frame and the statisticalresults of a given survey (what to do if the activity code of a unit obtained from a survey is different from the code in the sampling base from which the unit was sampled). There are divergent responses, however most respondents declare that they use the more up-to-date code in the processing of the survey. This is a major issue to discuss in the context of co-ordination economic statistics. The chosen solution can greatly affect the comparability of output of different statistics. In the questionnaire no distinction was made in real changes and non-real changes (corrections). It is a methodological issue how to deal with frame errors in survey procedures.

Major problems encountered

The responses for weak areas (question 4) and critical issues (question 5) can be dealt withtogether to provide a general view of the problems and challenges facing the SBR.

…