General preconditions

1
General preconditions

• Training workshop on censuses using
  administrative registers in Geneva 21 May 2012
• Harald Utne, Statistics Norway
• harald.utne_at_ssb.no

2
Preconditions for use of administrative sources
in production of statistics

• Legal base
• Public approval
• Unified identification system
• Comprehensive and reliable register system
• Cooperation among authorities

3
Legal base

• National legislation should reflect the idea of
  re-use of administrative data for statistical
  purposes
• Legislation should be up-to-date
• Statistics Act of Norway from 1989
• Replacing the former act of 1907
• Legislation must give the NSI
• Access to administrative data on a unit level,
  including identification information
• Right to link data from different administrative
  sources
• Legislation on data protection
• In Norway Principle of One way traffic
• Administrative data for statistical purposes
• Statistical data for
• Statistical purposes
• Research purposes (regulated by Statistics Act)
• Never for administrative purposes

4
Public approval

• Broad public approval is very important
• Influence on political decisions (legal base)
• Attitudes may be different between countries
  concerning
• Existence and linking of administrative registers
  in general
• Linking of register data for statistical
  purposes
• Lack of privacy NSI knows everything about
  everybody (Big Brother) or
• Data security better taken care of in centralized
  system
• Citizens may be reluctant to report data to the
  NSI if they know that the same data have already
  been collected for administrative purposes
• Common Nordic experiences
• Work of register authorities (and NSIs) should be
  open and transparent
• Open discussion and public debate

5
Unified identification system

• Unique identification numbers for basic units
• Persons, enterprises and establishments,
  numerical addresses
• Unique ID-numbers not absolutely necessary, but
  indeed very valuable
• Matching is possible without unique numbers, but
  more difficult and time consuming

6
Comprehensive and reliable register system

• Quality in administrative data important for
  individuals rights and duties
• Registers used by many authorities (incl. NSI)
  have best quality
• Information items that are important for
  administrative authorities, have the best quality
  
• But Administrative quality is not always the
  same as statistical quality

7
Cooperation among authorities

• Between different administrative authorities
  (registers owners)
• Between NSI and the registers owners
• Cooperation at all levels of the organisations
• Norway National forum of register owners and
  users
• The NSI needs a lot of information about the
  registers
• Regulations and practices in register keeping
• Contents
• Register owners should inform the NSI about all
  kind of changes
• In Norway regulated in the Statistics Act
• The NSI should inform the register owners about
  quality problems
• But normally not on an individual level

8
The Quality Circle
Public sector Statistics Private sector
Use
Work together
Quality
Updated information Quality secured Full
coverage Documentation
9
Requirements to data from administrative
sourcesAdditional requirements for producing
high quality register-based statistics
10
Units and identifiers

• Definition of units
• No problem for natural units (for example
  persons)
• Business units Must be defined by very precise
  rules
• New enterprise established or major changes in
  existing enterprise?
• Unique identifiers
• Identifying codes should not be changed during
  the lifetime of a unit
• Best solution Using codes containing no
  information
• Example Norwegian personal ID
• Contain date of birth and sex Changes may occur
• Temporary ID-number changed to permanent
  ID-number

11
Time references and data on events

• Important to have dates of changes and events
• Real point in time when possible Births, deaths
• Sometimes gradual changes Industry of an
  enterprise
• Dates of actual changes and dates of registration
• Distinguish real events from corrections in
  registers

12
Thank you for your attention !