14th International Roundtable on Business Survey Frames Auckland ; 29 October 3 November 2000

Transcription

1 14th International Roundtable on Business Survey Frames Auckland ; 29 October 3 November 2 Session No 6 Paper No 6 Susanne HAGENKORT, Federal Statistical Office, Germany Measures to improve the quality of the Statistical Business Register in Germany 1. Set-up of a quality policy Before setting up a quality policy, it is necessary to specify the term quality in the context of business registers. According to section 1 of the Manual of Recommendations on Business Registers, published by the Statistical Office of the European Communities (Eurostat), the quality of business registers is closely connected with user needs. Users want statistical registers to be relevant, accurate and up-to-date. Relevant means that the registers comprise the units and the variables attached to each unit necessary for the production of statistics. Details on what coverage, types of units and variables are considered to be necessary on the European level can be derived from the Council Regulation (EEC) No. 2186/93 of 22 July 1993 on Community co-ordination in drawing up business registers for statistical purposes (Register Regulation). Accuracy means that the information actually recorded corresponds to reality. Being up-to-date complements accuracy and implies that registers provide the most accurate picture of the real economic world with the least possible time lag. Upon knowing what determines the quality of registers, the next step in setting up a quality policy is to formulate and agree on certain quality standards. 1 These standards have to be realistic and achievable within a specified period of time. Thus, cost-benefit-arguments and given outline conditions must be taken into consideration. This will lead to the realization that a 1 % quality is clearly not the objective. But what is the objective, bearing in mind circumstances such as the initial stock of the register, the data availability, legal conditions, special national needs, and the demand to lower business response burden? The statistical offices of Germany have agreed on the objective of having available within the year 2 a register that contains all economically significant units from all branches of the economy in Germany, with data on annual turnover and number of employees. The attainment of this quality standard was supposed to be ensured through realizing the following milestones: 1) Processing the value-added tax files from the tax authorities for the first time, 2) processing the data file from the Federal Institute for Employment for the first time, 3) carrying out a register survey with the precise aims to delete double-recorded units, to shut down falsely active units and to store correct relations between different units. Each milestone was itemized into many consecutive steps, all laid down in a project and time schedule. The fulfillment of this schedule was supposed to be monitored at monthly intervals. The draft of the schedule was drawn up by a task force specially established for the so-called project Business Register System 95 (URS 95). It was agreed that this task force, which is managed by the Federal Statistical Office and composed of representatives in charge of business registers from both the Statistical Offices of the Länder and the Federal Statistical Office, should meet every two months to discuss and solve difficulties connected with the project. In the following, the implementation of the set up quality policy will be described. Particular emphasis will be put on the methodological-technical aspects of attaining each milestone and the effects of these different measures on the content and consequently on the quality of the business register, taking the initial situation into consideration. 1 For this paragraph see also Report of the Commission to the Council concerning the implementation of the Register Regulation, Commission of the European Communities, Brussels 2.

2 2. Implementation of the quality policy 2.1 Initial situation There did not exist a business register in Germany covering all sectors of economic activity when the Register Regulation was enacted in Within the scope of official statistics, however, several separate address files were available for individual sectors such as production industries and the sector of trade, hotels and restaurants. The first step in constructing a comprehensive business register in Germany was therefore the merger of the mentioned and some other available statistical address files into a register system to be known as the Business Register System 95 (URS 95). In accordance with the federal structure of Germany, this merger was carried out by the Statistical Offices of the Länder. At the end of 1998, the thus created business register contained the following number of active units subdivided by NACE-sections: Units of the Business Register (URS 95) at 12/1998 by NACE-Sections NACE-Sections Enterprises 12/1998 (Total: 2.2 million) Local units 12/1998 (Total: 2.4 million) A B C D E F G H I J K L M N O Agriculture, hunting and forestry Fishing Mining and quarrying Manufacturing Electricity, gas and water supply Construction Wholesale and retail trade; repair of motor vehicles, motorcycles and personal and household goods Hotels and restaurants Transport, storage and communication Financial intermediation Real estate, renting and business activities Public administration and defense, compulsory social security Education Health and social work Other community, social and personal service activities 4 2 A B C D E F G H I J K L M N O Federal Statistical Office of Germany, 14 th Roundtable, - I - Regarding the relevant sections according to the Register Regulation (C - K and M - O), two points have to be mentioned: first, units belonging to the sections C to H are already reflected quite well in 1998, if one compares the register world with the real world on the basis of other sources. Second, units active in the service sections I to K and M to O are evidently underrepresented compared to the real economic world. Thus, the gaps in the German statistical survey system become obvious. To fill these gaps in coverage and to obtain information on all characteristics demanded, it has been decided to use data already registered by other administrative bodies. The decision to use administrative data files was taken because in the medium term this is the only way in which the conflicting demands for more information about economic units on the one hand, and less business response burden on the other hand, as well as greater rationalization in official statistics can be reconciled. From a variety of administrative files four have been selected because they cover units from all economic sectors and contain information on characteristics considered important. To get access to these files, the bodies holding the administrative data had first to be legally obliged to make their data accessible for purposes of the statistical business register. The required national legal basis was established when the so-called Statistical Register Act was passed in This Act not only gives access to specific administrative files, but also allows the statistical offices to carry out inquiries to ensure correct linkages of units from different sources. After the legal conditions for implementing the above-described quality policy had been set up, some technical work had to be carried out to equip the register to accept administrative files. This was done by adding further characteristics to the data set, by extending the plausibility checks, and by adjusting the screen masks. After that, the first milestone, the processing of the value-added tax files for the first time, could be tackled. 2

3 2.2 Quality improvement through the use of the value-added tax files from the tax authorities The tax authorities value-added tax (VAT) files comprise all enterprises which are liable for VAT and have a certain size, namely a taxable annual turnover of at least DM 32 5 (= EUR ). Around 2.8 million of those enterprises and their annual turnover are covered. The processing of such a mass of data in the register requires in the first place the production of methodological-technical guidelines, also for the production of the necessary data processing programs. These guidelines were worked out by the Federal Statistical Office and agreed on in the task force URS 95. The programming was done in a statistical office that volunteered for this work as a member of a programming pool. The labor-intensive task of incorporating the VAT files into the register was the responsibility of the Statistical Offices of the Länder. This was done by preparing the files for the processing first, e.g. by checking them for plausibility, by searching for identical units from the tax files in the register next, and by inserting characteristics from the VAT files into identical register units as well as transferring units not found from the VAT files to the register afterwards. The following diagram shows how these activities have affected the coverage of the register: Enterprises of the Business Register (URS 95) before and after the processing of the VAT files by NACE-sections NACE-Sections Enterprises 12/1998 (Total: 2.2 million) Enterprises 3/2 (Total: 3.5 million) A B C D E F G H I J K L M N O Agriculture, hunting and forestry Fishing Mining and quarrying Manufacturing Electricity, gas and water supply Construction Wholesale and retail trade; repair of motor vehicles, motorcycles and personal and household goods Hotels and restaurants Transport, storage and communication Financial intermediation Real estate, renting and business activities Public administration and defense, compulsory social security Education Health and social work Other community, social and personal service activities 4 2 A B C D E F G H I J K L M N O Federal Statistical Office of Germany, 14 th Roundtable, - II - The diagram illustrates clearly that the use of the VAT files has led to an acquisition of extra units, in particular of enterprises which come under the service sections not nearly covered appropriately before. Besides, the variable turnover has been completed. Therefore, a very important aimed at quality standard has been achieved through processing the VAT files in the register. The achievement of this first major aim, however, also leaves something to be desired with regard to the quality of the register. For example, the relatively poor increase of enterprises belonging to the NACE-sections M and N is due to the fact that enterprises active in these sections predominantly either do not make any turnover at all or only turnover that is not subject to VAT. The same partly applies to units in section J, e.g. units with activities auxiliary to insurance. Thus, the units concerned are still underrepresented in the register after having processed the VAT files. Furthermore, the definition of the unit enterprise according to VAT law in principle tallies with the respective definition used in federal statistics, but there is one exception. Legal persons who count as independent enterprises for the purpose of federal statistics are not considered to be independent from the point of view of VAT legislation if, as controlled subsidiary companies [ Organgesellschaften ] they are assigned to an affiliated group of companies [ Organschaft ]. From the tax legislation point of view, the affiliated group together forms a single enterprise consisting of the parent company [ Organträger ] and controlled subsidiaries. Data on the subsidiary companies are therefore not included in the VAT files. The taxable turnover of these units is shown together with that of their parent company. 3

4 With regard to the business register this means, first, that the subsidiary companies which were not covered by the register before processing the VAT files are still not covered at this stage, second, that for the subsidiary companies already stored in the register no taxable turnover is shown, and third, that the taxable turnover registered for the parent company refers to the affiliated group. Another very important reason for the limited quality of the register, although the VAT files have been processed, is that identical units from the tax files and the register could only be matched via addresses and not via universally valid codes stored in the two sets of data for a given unit. Even though a software was used to support the identification of units via addresses, this method is not only immensely labor-intensive, but also often unsuccessful. The tax authorities files frequently include addresses used for the dispatch of tax bills, i.e. addresses at which the enterprises owners reside or addresses of other authorized persons such as tax advisers. This means that the software fails to recognize units as being identical if the address in the register is that of the residence of the enterprise. Even with a good deal of manual research to establish the identity of units, the duplication of units already stored in the register could therefore not be prevented completely when processing the VAT files. Aware of the quality standard achieved through the first use of the VAT files, the implementation of the second milestone, the processing of the file from the Federal Institute for Employment for the first time, was taken on. 2.3 Quality improvement through the use of the file from the Federal Institute for Employment (FIE) The FIE s file comprises all local units with employees who are liable to pay social insurance contributions. These data, covering around 2 million units and the number of their employees, were integrated into the business register in a similar way as the VAT files. After preparing the file for the processing, identical units from the file were searched for in the register that had already been expanded through the data from the VAT files. If a match was found, certain characteristics were inserted into the register unit. If no match was found, the unit concerned was transferred to the register as an extra unit. Extra units had to be inserted as enterprises instead of local units, because the FIE does not provide any information on whether a local unit is an enterprise itself or a part thereof, and if a part thereof, to which enterprise the local unit belongs. The following diagram shows how the coverage of the register has been affected by the processing of the data file from the FIE: Enterprises of the Business Register (URS 95) before and after the processing of the file from the Federal Institute for Employment by NACE-sections Enterprises 3/2 (Total: 3.5 million) Enterprises 7/2 (Total: 4.1million, partly estimated) NACE-Sectio ns A A gricultur e, hunting and fore stry B F is hi ng C Mining an d quarry ing D Manufactu ring E Electricity, gas and water supp ly F C ons truction G W ho les ale an d retail trad e; repair o f motor v ehicles, moto rcycles an d pers ona l and househ old go od s H H otels an d resta uran ts I Tran sport, storage an d commu nication J Finan cial in termed iation K R eal estate, renting and bu siness activities L Pu blic ad ministr ation and d efens e, c omp ulsory socia l secu rity M Ed ucation N H ealth an d so cial wor k O O ther co mmunity, social an d per sonal s ervice activitie s 2 A B C D E F G H I J K L M N O Federal Statistical Office of Germany, 14 th Roundtable, - III - 4

5 The use of the data obviously has led to an acquisition of extra units in all NACE-sections and in particular of units which come under the service sections I, J, K, M, N and O. The above-described quality defect that related to the coverage of units belonging to the sections J, M and N thus has at least been reduced through processing the FIE s file. The same is true for the lack of information on the existence of subsidiary companies. Since the local units of almost all controlled subsidiary companies have employees liable to pay social insurance contributions, the statistical offices have at least obtained information on the existence of the companies concerned. Besides, the variable number of employees has been completed. Bearing all this in mind, a further, higher quality standard has been achieved through processing the FIE s data in the business register. But again, the achievement of this second major goal leaves something to be desired. For example, units with no employees liable to pay social insurance contributions are not included in the file from the FIE. For instance, data on self-employed businessmen or the self-employed working in the liberal professions with only unpaid family workers or persons in minor employment are therefore not transmitted by the FIE. If no other source provides information on these units, they are underrepresented in the register. Since the VAT files contain information on enterprises regardless of whether they have employees, this quality limitation has been assessed as being of less importance on the level of enterprises. It is rather assumed that the very vast majority of economically significant enterprises is covered by the register now that the VAT files and the FIE s file have been incorporated. On the level of local units, apart from the units that have no employees liable to pay social insurance contributions, further units are missing in the FIE s data. This is due to the fact that the delimitation of local units according to the FIE does not in all cases correspond to the one which has to be applied in the business register. Although the FIE also uses in principle a geographical criterion to delimitate local units, there is one exception. Local units which belong to the same enterprise, are located in the same municipality [ Gemeinde ] and have the same activity-code are generally combined into a unit known as the master establishment. Data on local units belonging to a master establishment are therefore not included in the FIE s data. Their number of employees is shown with that of their master establishment. With regard to the business register this means, first, that local units which either have no employees liable to pay social insurance contributions or belong to a master establishment are not covered by the register, if they were not covered before processing the file from the FIE. Second, for the units stored in the register and belonging to a master establishment no number of employees is shown. Third, the number of employees registered for the local unit representing the master establishment refers to the whole master establishment and not the local unit concerned. As has already been mentioned, another reason for the limited quality of the register, even after the FIE s file has been processed, is that the file provides no information on whether a transmitted local unit forms an enterprise itself or just a part thereof, and if a part thereof, to which enterprise the local unit belongs. Consequently, units not matched with a register unit had to be added to the register as enterprises regardless of whether they actually are an enterprise themselves or just a part thereof. In the case of added units forming part of an enterprise in the real world, the type of unit stored in the register does not correspond to reality. Furthermore, the reference to the real enterprise is missing in these cases. Apart from the quality defect resulting from an insufficient reflection of relations between different types of units, the duplication of units again could not be prevented completely when processing the file from the FIE. This is due to the difficulties caused by the matching of units via addresses, as described above. In view of the quality standard achieved through the first use of the VAT files and the file from the FIE, the third milestone, the carrying out of a register survey, was tackled. 2.4 Quality improvement through a register survey The register survey started in June 2 and its largest part will probably be finished towards the end of November 2. The main aim of the survey is to remedy the most serious quality defects of the register existing on the level of enterprises. The decision to focus on the level of enterprises was taken because, among other things, reliable statistical data on enterprises are in stronger demand by users than the corresponding data on local units. 5

6 The precise aims of the survey are 1) to delete double-recorded units which have been caused primarily by the use of administrative files, 2) to shut down falsely active units which have been caused mainly by an insufficient maintenance of the initial business register, 3) to store correct relations between different units, i.e. between controlled subsidiary companies and their parent company as well as between local units and their enterprise. In a first phase, it had to be clarified which units should be included in the survey. Two points had to be taken into account: on the one hand, that the targeted quality standard can be reached, at least from a theoretical point of view, and on the other hand, that the number of units questioned should be minimized as far as possible. Since the quality improvement is to focus on the level of enterprises, the number of units to be questioned could first be limited to those registered as enterprises. Further limitations were dependent on the source from which these units originate. At present, the register contains units which essentially come from three different sources: statistical address files that had been available before processing any administrative file, the VAT files and the file from the FIE. Depending on these sources, the following cases of doubt were found to require clarification through the survey: 1) Enterprises which come from statistical address files and could not be matched with a unit from the VAT files. These units may not have been found in the VAT files, because of divergent addresses or because they are not contained in the VAT files. In the first case, the units are double-recorded and should be deleted by asking them for their identification code in the VAT files. In the second case, the units concerned may have run out of business in the real world and should therefore be shut down in the register, too. Another possibility is that the units concerned are controlled subsidiary companies and should then be asked for information on their respective parent company. 2) Units which come from the FIE s file and could not be matched with any unit stored in the register. These units may be double-recorded or subsidiary companies, too. Furthermore, it is possible that they are equivalent to a local unit which is only part of an enterprise in reality. In this case, the unit concerned should be asked for information on the respective enterprise. Having determined which units should be included in the survey, the next step was to define what the units should be asked about in detail. This had to be done by taking into account, again, that the targeted results should be achieved with the least possible response burden. Besides, every question had to be covered by the provisions of the Statistical Register Act. Consequently, the questionnaire developed consists of one page only (see appendix). The following information is requested from the units: 1) Information on whether they are still economically active or not. Through this question it will be possible to clean the register from deadwood, i.e. units who have run out of business but are still characterized as being active in the register. 2) Corrections of their address, if necessary. This is a very useful side effect of the survey, because it ensures the registration of correct addresses for units questioned. 3) Information on whether they consider themselves as being an enterprise or just a local unit. This allows to store the correct types of units in the register. 4) Information on their identification codes in the various administrative files to be used for the purpose of the register. Through these data it will be possible to delete double-recorded units because identical units can be matched clearly using identification codes. Besides, it will be easier to identify units in administrative files still to be used for the purpose of the register. Thus, the difficulties caused by matching the units via addresses will be reduced in the future. 5) Information on whether the unit is a controlled subsidiary company, and if so, the address and tax identification code of the respective parent company (only requested from enterprises). This question allows to store correct relations between affiliated groups. 6

7 6) Address and tax identification codes of the respective enterprise (only requested from local units). This information makes it possible to reflect correct relations between local units and their enterprises in the register. In a next step, it had to be decided how the information obtained through the survey should be processed in the register in detail. Some methodological-technical guidelines had to be developed. Furthermore, a data processing program was needed, so that at least some of the processing-work can be done computer-aided. Upon having established the methodological-technical preconditions, the Statistical Offices of the Länder forwarded the questionnaire to 1.2 million units. Since the inquiry is still going on at the moment, definite results are not available yet. Very first experiences, however, give reason to be confident that the targeted quality standard is not only achievable from a theoretical point of view, but can also largely be put into practice, even if this takes a lot of time because most of the work has to be done manually. 3. Assessment and outlook The statistical offices of Germany set up a quality policy with the objective of having available within the year 2 a register that contains all economically significant units from all branches of the economy, with data on annual turnover and number of employees. On the level of enterprises, it is assumed that the objective will be achieved, after processing the VAT files, the file from the FIE and the information obtained through the register survey. Only those enterprises which are neither liable for VAT nor have employees liable to pay social insurance contributions and were not covered before the processing of these administrative files will still be missing in the register. Furthermore, for those enterprises that are neither liable for VAT nor covered by any statistical survey, the variable annual turnover will not be available. On the level of local units, those will still be missing that were not covered before processing the file from the FIE and either have no employees liable to pay social insurance contributions or belong to a master establishment. Altogether, the quality standard that is expected to be achieved through the realization of the three milestones described above is assessed as remarkable progress. The strict monitoring of the project and time schedule as well as the meetings of the task force at regular intervals to discuss and solve difficulties early have contributed to achieve this standard and thus have proven to be a success, too. For the future, it is intended to achieve - step-by-step - higher quality standards. For example, it is planned to process the income and corporation tax files to gain information on enterprises not covered in the register yet as well as to gain information on the size of enterprises, even if they are not included in the VAT files. Another step towards improving the quality of the business register will be to incorporate information from the files of the Chambers of Commerce and Industry into the register. These files, among other things, contain data on local units, even if the units have no employees or belong to a master establishment. Furthermore, detailed concepts have already been developed on how to update the register with the VAT files and the file from the FIE. Since the updating can be carried out faster than the first processing, the register will be more up-to-date than at the moment. In the course of time, all these measures will contribute to meet the user needs better and better and thus will improve the quality of the business register. 7

8 APPENDIX - Questionnaire used for the register survey - Statistisc he s Landesamt Registerumfrage Identitäts-Nr.: Rücksendung bis zum XX.XX.XXXX Die Rechtsgrundlage finden Sie auf der Rückseite des Erhebungsvordrucks Für Ihre Fragen wählen Sie bitte: Telefon: () - XXX XXX XXX SST 1-9 (bei R ückfragen bitte angeben) XXXXX XXXXXXXXXXXXXXXXXXXXXXXXXXX Berichtigen Sie bitte ggf. Name, Anschrift und Rechtsform (Firmenstempel): Die vorgenannte Einheit ist 1 1 wirtschaftlich tätig. 1 wirtschaftlich nicht mehr täti g. (Bitte senden Sie den Erhebungsvordruck ohne weitere Eintragungen zurück.) Handelt es sich um ein Unternehmen oder eine Zweigniederlassung? Bitte kreuzen Sie entsprechend an und tragen Sie die für die Einheit zutreffenden Angaben ein. Erläuterungen zu den von Ihnen einzutragenden Angaben finden Sie auf dem Beiblatt Ausfüllanleitung. 1 Unternehmen! 2 Zweigniederlassung " Angaben zum vorgenannten Unternehmen: Betriebsnummer bei der Bundesanstalt für Arbeit Mitgliedsnummer bei der Industrie- und Handelskammer ODER Angaben zur vorgenannten Zweigniederlassung: Betriebsnummer bei der Bundesanstalt für Arbeit 2 32 Mitgliedsnummer bei der Handwerkskammer Mitgliedsnummer bei der Industrie- und Handelskammer Steuernummer (für die Veranlagung zur Umsatzsteuer) 46 Umsatzsteuer-Identifikationsnummer (USt-IdNr.) D E Ist dieses Unternehmen eine Or gangesellschaft einer Umsatzsteuer-Organschaft? 1 7 ja 7 # nein, bitte weiter auf der Rückseite 58 Angaben zum Unternehmen dieser Zweigniederlassung: Name und Anschrift des Unternehmens Angaben zum Organträger dieser Organgesellschaft: Name und Anschrift des Organträgers Steuernummer des Unternehmens (für die Veranlagung zur Umsatzsteuer) Umsatzsteuer-Identifikationsnummer (USt-IdNr.) des Unternehmens D E Steuernummer des Organträgers bitte wenden 8