Reengineering the French Business Register (2001-2006)

1. The context: an inter-administrative BR, core of the statistical BR system

As this is the case for several countries, the French Business Register underwent recently a major re-engineering. The aim of this presentation is to provide its main characteristics.

It must be kept in mind that the structure and operations of a Business Register is dependant of the national context. In France, INSEE is, by law (decree of 1973), in charge of a unique inter-administrative business register (SIRENE): the use of its identification number has been made mandatory for relations between businesses and administrative bodies (Tax administration, Social security, etc.) by a Government’s decree of 1997. The information contained in this inter-administrative register is not legally binding, but its identification number is also used in legal registers.

New registrations (i. e. creations) and updates are transmitted to INSEE on a daily basis through so called CFE’s (Centres for Business Procedures) managed by Chambers of Commerce, of Agriculture, of Handicraft, etc. According to its status and/or industry, each business is allocated to one CFE, which is able to process its declarations and re-send data to INSEE and to each relevant administrative body.

This inter-administrative BR (SIRENE) is of course the core of the statistical BR. Since the beginning INSEE took great care to have SIRENE be at the same time an identification system and a permanent census of business units; this was to be obtained by:

-aiming at generalizing the use of the unique identification number (which succeeded in 1997)

-improving identification techniques

-increasing the number and the quality of variables, in particular by using information from statistical surveys, either through a Law authorizing such use, when it is considered useful to the public (e. g. industry codes, size classes, …) or by adding variables the use of which is restricted to statisticians.

The re-engineering of SIRENE then aimed at improving both inter-administrative and statistical dimensions of the system:

-re-enforce the central role of INSEE within the inter-administrative system

-re-enforce the role of SIRENE within the business statistics system

-and, last but not least, improve efficiency and reduce costs

The process involved a review of the register itself, i.e. units and variables, of software used for managing and accessing data, and of the computer architecture. These three components were impacted in the operation, which was first delineated in April 2000 and should be completed by mid 2006.

It must be stressed that SIRENE is only a part of the statistical BR system: the French business statistics system itself is currently also undergoing a re-engineering operation, called RESANE, which will impact the other components of the statistical BR. One of its main features is to consider the enterprise groups as a major observation unit for economic statistics. The re-engineering of SIRENE is then the first - and hopefully major step of the re-engineering of the French statistical BR. This paper will stress on the role of SIRENE in the statistical BR system and on the improvements in this particular domain due to the re-engineering process of 2000-2006.

2. SIRENE and the statistical BR before the re-engineering

Until now, the only – but essential – feature of SIRENE for contributing to the statistical BR was the identification and the management of legal and local legal units in the scope of business statistics, with the addition of a few ad-hoc variables.

2.1 Keeping track of the birth and administrative life of enterprises

Managing the inter-administrative register implies the recording and identification of the units, and the taking into account, on a daily basis, of about 2million movements per year (creation, updating of various variables, deletion). As a result of the central role of the register and of time constraints due to government initiatives aimed at developing entrepreneurship, INSEE is made aware of any birth of legal or local legal unit within a very short notice.

2.2 Delineating the field of business statistics

Codes are allocated in order to permit the marking of legal or local legal units, which have an interest as far as business statistics are considered. Among the 6.5 million legal units of the register, 4 million legal units have an economic interest, many of which may be considered as “enterprises”.

2.3 Managing an economic activity status

The inter-administrative BR in itself can only record the administrative status of a unit, i. e. existing or ceased. Termination of businesses are known with delay: 78% after 3 months, 88% after 6 months, 94% after one year, 97% after two years, about 100% after three years only. On the other hand the inter-administrative BR must, by law, reflect the existing Legal Registers, in particular as to the legal existence of a unit.

For this reason, a variable “economic termination” is introduced, for statistical purpose, indicating the end or the continuation of the economic activity of a unit, independently of its legal situation. The relevant information is obtained from the processing of statistical surveys or from the carrying out of specific “BR improvement surveys”.

2.4 Additional features

Economic continuity of local units is characterized by the use of a specific kind of unit (ETEC)

Codes describing institutional sectors are allocated to units for the needs of National accounts.

Another useful characteristic of the inter-administrative register is the identification of legal units, which have no establishment in France but have to pay social contributions and/or taxes. Their number is increasing due to new regulations. SIRENE is then able to identify units outside the country for other purposes, e g in order to register, for statistical needs, the components of multinational enterprise groups.

3. Main improvements from the re-engineering

The re-engineered inter-administrative BR register (SIRENE3) brings many improvements to the statistical BR system of INSEE.

3.1 A new architecture, allowing the management of new kind of units

Even if 90% to 95% of economically active enterprises - in numbers, if not in economic weight - correspond to one legal unit, the need of identifying other units is more and more urgent in business statistics.

It will be possible with SIRENE3 to manage links of any type (financial, network, franchise, co-operation,...) between legal units and between local legal units in order to delineate a perimeter for more complex units. The record of a unit may contain references to specific tables, one for each kind of link. These tables contain the list of linked units, with variables qualifying the link (for example beginning and end dates). The list of characteristics in the table depends on the kind of link the table describes.

Three kinds of links are, for the time being, possible to be managed on a regular basis: financial links (ownership) between legal units, other links between legal units (e g association of professionals (doctors, lawyers..) working together, or sharing an office or staff), links between local units dependent of several legal units (staff lending, franchise…). Other kind of links will be defined when a regular updating source is found.

Moreover, new kind of units may be defined, in addition to the legal and local legal units of the inter-administrative register, with their own set of variables (industry code, size…), allowing the identification and description of complex enterprises, in addition to the delineation of their perimeter by the use of links..

As the architecture of the new system, provides statisticians with specific working stations, they will be able to manage complex units like enterprise groups, on the same way the staff in charge of the inter-administrative register manages legal and local legal units.

3.2 New variables for the statistical BR

3.2.1 Economic activity

As said before (2.3), an “economic termination” code exists in SIRENE for statistical purposes. With the reengineering two new features are introduced, for use by official statisticians only:

• a “termination assumption” code, triggered by any reliable information source, without waiting the result of any formal survey

• for any unit said to be active in the Register, a probability ratio to be effectively economically active at the end of the year or at a given date, mainly derived from the date of the last known event concerning this unit.

This last feature allows for an estimation of the stock of economically active businesses in a given population. It may also be used for launching specific “BR improvement surveys” with a higher efficiency, which is a way to reduce the cost of maintaining the quality of the BR.

3.2.2 Economic continuity

SIRENE3 allows for a comparison between successive periods by managing links between predecessors and successors for each kind of unit in the register: an important feature of this link is economic continuity.

Statistical users will be provided with annual and quarterly files with predecessor-successor tables, in order to allow the use of administrative sources in which updates of identifiers are often made with an important delay.

3.2.3. A background history for the main variables.

For the main variables - codes defining different economic and legal status, industry code, number of employees, turnover, identifier of headquarters - successive values, together with dates of processing and dates of effect are recorded in SIRENE3. At the launching of the new database the background history will be from 2 to 5 years old, according to variables. This should be considerably helpful for producing annual reference files and for some statistical studies.

3.2.4. A comprehensive and synthetic information on each unit

With more kind of units - and a flexibility for adding new ones, more variables per unit - with the same flexibility, links, the content of SIRENE is far richer than before. Its function is however strictly kept to that of a register, not a database. It means that its variables are only for identification and classification purposes, i.e. that SIRENE is not a repository of economic data used for computing statistical results, other than business demographic statistics.

The list of variables contributing to the statistical BR is the result of a dialogue within the statistical system, some variables are issued from the inter-administrative management, other from various data sources, surveys of administrative files. For each variable a unique reference source has been chosen.

Variables may be classified in:

-identification variables (e g identifier in a European register)

-status variables (administrative and economic, legal situation…)

-classification variables (industry code, institutional sector, number of employees, turnover…)

-variables used in enterprise demography

-link variables

besides variables used for managing the register (e g access of different kind of users).

3.3 SIRENE3 allows independent operations related to the management of the statistical BR

Notwithstanding that the inter-administrative BR and the statistical BR are two different concepts, it was decided to take both aspects into consideration while building the new inter-administrative BR. Instead of building two registers the choice was made:

• to have a single register
• to distinguish between operations concerning the management of the inter-administrative register, the management of the statistical BR, or both aspects
• to allow a degree of independence in the management of the different aspects.

Besides variables or modalities specifically used for statistics (e g economic termination, institutional sector..), besides the use of various sources outside the inter-administrative network, some technical choices are aimed at making statistical management independent from administrative management:

3.3.1 An entirely automatic coding for variables of statistical use

Variables used only for statistical purposes (industrial classification, institutional sector,…)are defined with their users -statisticians, national accountants..- and their definition is kept up to date and available in a documentary database (SYDORE). They are automatically encoded when the inter-administrative register is updated, ensuring completeness and homogeneity of the register for its use within the statistical BR system.

3.3.2 Introducing specific events for amending errors

This new feature allows drawing a distinction, among updates of a given unit, between “genuine” updates and error corrections. These corrections may result from the inter-administrative management as well as from quality controls by INSEE.

3.3.3 Enabling to process “statistical” events as well as administrative events

The architecture of SIRENE3 allows to process in a similar way events from the inter-administrative area (so-called CFE - cf. supra) and events from the statistical area (e g updates resulting from quality campaigns). Any update of the register is made through “bundles” describing “administrative” or “statistical” events using the XML language.

For processing individual events (error corrections, new information on a unit…) the modularly design of SIRENE3 allows an easy development of dedicated workstations for staff in charge of statistical operations.

For mass operations dealing with the same event applied to many units, flows of bundles may be issued by statistical processing systems.

The flow of bundles from statistical operations, from dedicated workstations, from the inter-administrative area, i. e. statistical as well as administrative events are, as the case may be, automatically processed or put forward to the workstation of the staff in charge of managing the BR. Priorities may be assigned in order to process the various kind of bundles on a flexible way.

3.3.4 Taking into account the fact that administrative workflow and statistical workflow may be desynchronized

The re-engineering of the BR makes the coding of variables in the statistical BR independent of the administrative process: this important feature allows to postpone the statistical update until all necessary information is available and validated without impairing the quickness of administrative processing.

It will, for example, be possible to register a new business and send its identification to its “CFE” and to administrative partners even if some statistical variables - for example the “predecessor” identification - require further investigation to be coded. In the past, the pressure of administrative management could result in a loss of quality for statistical purposes.

The new architecture also allows triggering any kind of quality control on the units already in the BR, independently of the current flow of process. Errors or inconsistencies may be redirected to specialized staff in order to update the BR, giving the possibility to redefine any variable of statistical interest.

The management of the background history of variables allows to insert values for past time periods. It will then be possible to record information when it is available, without consideration of delay. For example, updating the industry code of a unit from the result of a statistical survey for a given period will be possible regardless of its actual industry.