Business Registers Recommendations Manual.ppt

Document Sample
Business Registers Recommendations Manual.ppt Powered By Docstoc
					Data sources of the
EuroGroups Register
Presentation by Eurostat
Zsolt.Volfinger@ec.europa.eu
Scope of the presentation

 Scope of the presentation:
   – What is the EuroGroups Register (EGR)?
   – Data acquisition.
   – The commercial data sources of the EGR.
   – The composition of the starting population.
   – The characteristics and the completeness of the initial
     datasets.
   – The results of data exchange with EU and EFTA Member
     States.
The EuroGroups Register

 The EuroGroups Register is the statistical register of the
  European Communities on multinational enterprise groups
  (MNEs).
 The aim of the register is to cover the units of the most
  influential MNEs operating in the EU and EFTA countries to
  create a harmonised frame for European statistics on
  multinationals.
The EuroGroups Register

 The EGR is based on data of two commercial data sources.
 The information of these sources is combined within the EGR.
 The national statistical institutes (NSIs) of the EU and EFTA
  countries are involved in the EGR production process.
  Feedback from the actions of the NSIs:
   – The NSIs link the records of the commercial data sources
     to their national business registers.
   – The NSIs check and correct the characteristics and the
     relationships of the units at record level.
 The final MNE structures will be defined based on the parent-
  subsidiary relationships by the EGR system.
 EGR will disseminate the data of MNEs to Member States.
Data acquisition

 In 2008 Eurostat announced a tender for private data
  providers to supply data on multinational groups for statistical
  purposes to the EuroGroups Register. The tender was split to
  two lots. One lot requested data on control relationships, the
  other one data on ownership relationships.
 The tender specification required data of 5000 MNEs with the
  characteristics of all involved subsidiaries and the relations
  between them.
 For the dataset on control relationships the winner of the
  tender was company Dun & Bradstreet (D&B), for the dataset
  on ownership relationships company Bureau van Dijk (BvD).
 Both of these private data providers do have a worldwide
  commercial database on multinational groups.
Selection procedure for groups

 Starting population: 5000 MNEs and their legal units.
 Coverage: to cover the most influential MNEs active in
  Europe, those which are the most relevant from the
  perspectives of FATS statistics, FDI statistics and
  employment.
 Every EU and EFTA country are represented in the 5000
  population.
Selection procedure for groups

 Country lists were requested from EU and EFTA Member
  States on influential MNEs which are resident in their country.
 The number of requested MNEs by countries was predefined
  by Eurostat.
 Member States provided 3000 supposed MNEs for Eurotsat.
 D&B could match 2410 groups to its database; BvD could
  match 2729 groups. These matching units composed the first
  half of the starting population.
 Next step: country specific selection from the private providers
  for the second half of the starting population.
 Additional non country related MNEs from the data providers
  to top up the 5000 starting population.
Data deliveries

 Data delivery from private sources in January 2009.
 Time stamp of the datasets: end 2008.
 D&B: 5000 MNEs and their units in 330 296 records.
 BvD: 5452 MNEs with 505 560 records and a separate file on
  ownership relations between the records.
 Data delivery in text files.
 The final delivery files included information on the
  characteristics of the legal units of the MNEs and on their
  relationships. Both sources provided more than 50 variables
  for the records.
D&B dataset

 Variables             Filled records     %
 DUNS number                 330 296    100%
 National ID number          265 704    80%

 The national ID numbers are very important for the linking
  process.
 The data file had a good coverage on identification numbers
  80% of the records did have national ID number and only
  1.8% was missing for European records.
D&B dataset

 Variables             Filled records     %
 Company name                330 296    100%
 Country code                330 296    100%
 Address                     324 975    98%
 Legal form                  300 440    91%

 Every record had a company name.
 The country and country code were filled by every record.
 5300 records did not have address line, 3000 did not have
  city name.
 91% of the records did have D&B legal form.
D&B dataset

 Variables                     Filled records     %
 NACE Rev 1.1. activity code        308 061     100%
 D&B estimated employment           330 296     100%

 At least one NACE Rev 1.1 code was available by 91% of
  the records. We faced a quite high number of invalid NACE
  codes, 22% of the delivered codes were not valid according
  to the NACE Rev 1.1 classification.
 D&B delivered an estimated variable for employment for
  every record. Turnover and net worth estimations were also
  available for every record.
D&B data on relationships and summary

 The D&B database is based on control relationships without %
  data.
 Natural person are not included in the group structures, only
  legal units are covered.
 Every record had an immediate parent link with DUNS
  number. The domestic and global ultimate units were defined
  by every record.
 The relationship data between the records were very clear
  and consistent; this was a strong side of the D&B dataset.
 Good coverage on ID numbers, names and addresses, on
  legal forms, on employment, turnover and net worth figures
  and on links between records.
 The coverage for Greece is poor; D&B could provide data only
  for 2 influential Greek groups.
BvD data file on legal units

 Variables              Filled records     %
 BvD number                  505 560     100%
 National ID number          368 343     73%

 73% of the total population did have national ID number in
  the BvD database.
 12% of the records did have a "DNB company number",
  which is the DUNS number of the records from Dun &
  Bradstreet. These numbers gave additional possibilities to
  link the data of the two sources.
BvD data file on legal units

 Variables             Filled records     %
 Company name                505 560    100%
 Country code                495 635    98%
 Address                     351 282    69%

 Every record had a company name.
 2% of the records did not have country code; this is a
  relatively high number. The EGR can not use records
  without country code; these records are excluded from the
  MNE structures.
 The street address was available by 69% of the records,
  75% did have a city name.
BvD data file on legal units

 Variables                     Filled records    %
 Legal form                         312 864     62%
 NACE Rev 1.1. activity code        312 120     62%
 Employment                         198 970     39%

 62% of the records did have legal form. These records were
  described by 256 different legal forms.
 62% of the records did have NACE Rev 1.1 code. 3% of the
  codes were invalid according to NACE Rev.1.1
  classification.
 For 39% of the records employment figure was available.
BvD data file on relationships

 The relationships between legal units were delivered in a separate
  file.
 Contrary to the D&B dataset minority ownership relations as above
  10% are included as well.
 Private persons are covered as group heads.
 The relationships are described by direct or by indirect percentage
  values. BvD provided % value for every relation. 85% of the relations
  are direct.
 If the direct percentage was empty an indirect (total) percentage was
  always available. Time stamp and source description was available
  for every record.
 In some cases the file on linkages from BvD included controversial
  information. It happened that for the same legal unit two different
  parent companies were recorded, both with 100% ownership.
Summary on the BvD dataset

 The BvD database includes % values for all relationships.
 Majority relationships and minority ownership relations as
  above 10% are included as well.
 Outside the legal units natural persons are covered as well.
 The identification of natural persons is difficult; it happened
  that they do not have country codes, the have only a name
  and a BvD number.
 The BvD dataset had a good coverage on names and
  addresses and on links between records.
 The availability of the national IDs and of the activity
  classification was limited (73% and 62%).
 2% of the records did not have country code.
Summary on the BvD dataset

 Data from BvD usually includes more records for the same
  group than D&B data.
 The BvD data set includes a larger proportion of local units
  and ceased units.
 It seemed that BvD tried to collect every piece of information,
  e.g. from administrative registers, annual accounts, etc, and in
  some cases they stored different information on the same
  relationship.
 The data were not every time consistent, but these were very
  few exceptional cases.
1st data exchange with Member States

 In January 2009 the legal units of the two sources were linked
  at record level by automatic processes of the EGR.
 In February 2009 Eurostat completed the first data exchange
  with the Member States; the NSIs received all linked legal
  units resident in their territory.
 The NSIs examined the data sets; they linked the records of
  the data providers to their national statistical business
  registers, they provided them proper national identification
  numbers.
 The results of the first data exchange gave information on the
  quality of the national ID numbers in the source data sets.
 From the 31 EU and EFTA Member States 29 countries
  cooperated in this action.
1st data exchange with Member States

 From the D&B data 235 757 European records were sent to EU and
  EFTA countries to link to the national business registers.
 4400 records did not have national IDs (1.8%).
 The quality of the national ID numbers was good, the NSIs could link
  91.2% of the D&B records successfully to their national registers.
 They accepted further 5%, these records are national records, but
  because of register thresholds they are not included in the statistical
  business register.
 1.3% of the ID numbers had to be changed, 1.5% was added.
 Only 1% of the provided records remained unknown.
 99% of the D&B records are real entities; they are stored in the
  national business registers.
1st data exchange with Member States

 From the BvD data 344 821 records were sent to Member
  States to link to the national business registers.
 42 000 records did not have national IDs (12.2%). The
  countries could link 80.9% of the BvD records successfully to
  their national registers.
 They accepted further 3.7%, these records are national
  records, but because of thresholds they are not included in the
  statistical business register.
 2.5% of the ID numbers had to be changed, 7.6% was added.
 5.3% of the provided records remained unknown.
2nd data exchange with Member States

 Based on the linked national ID numbers the EGR created
  legal units linking the BvD and DnB records.
 In case of conflicting information the system gives priority for
  DnB.
 The characteristics of these records and the relationships
  between the units are checked and corrected by EU and
  EFTA Member States as well.
 This was the task of the second data exchange in the period
  July-August 2009.
 Up to the end of September Eurostat has received 21
  datasets from the Member States.
2nd data exchange with Member States

 Feedback from Member States on status of the legal units
 Status of legal units in the answers:
   – Active             86.7%
   – Liquidated          7.6%
   – Inactive            1.8%
   – Unknown             3.9%
2nd data exchange with Member States

 Feedback from Member States on status of relationships
  between legal units
 Status of relationships in the answers:
   – Active                     65.6%
   – Ceased                      9.2%
   – Wrong, never existed        6.8%
   – Unknown, not confirmed 18.4%

				
handongqp handongqp
About