1 Administrative sources for official statistics why

1 Administrative sources for official statistics – why and how? Øystein Olsen, Director general Statistics Norway 1

Some issues to be addressed: • What is administrative registers and official statistics? • Why use administrative sources for official statistics? • How use administrative sources for official statistics? 2

The main differences between administrative registers and official statistics • Administrative sources/registers: – Normally based on a specific legal act regulating purpose and use – To be used for administrative purposes – also for handling individual cases – Should cover the total target population with good quality and updating of important variables/indicators • Official statistics: – Based on a legal act limiting i. a external access to microdata – The focus is analysing/disseminating aggregated/statistical information of public interest – Based on sample surveys, censuses and administrative data, often at micro/individual level 3

Why use administrative sources for official statistics • Cost efficiency – more statistics with less costs • Reduce response burden • Avoid overlap in data collection • Normally good coverage, quality and updating of basic information 4

General preconditions for producing statistics based on administrative data • Legal base Statistics act and Act on processing of personal data – Gives SN legal access to administrative data w On a unit level w Linking data for statistical purposes – Data protection w “One way traffic”: Administrative data for statistical purposes, not statistical data for administrative purposes • Public approval – The general public opinion in Norway: w Statistical use of administrative data is rational w Strong faith to SN (reliable, independent) – Open discussion and public debate important 5

General preconditions (cont. ) • Unified identification code system – Identification numbers for all basic units: Persons, dwellings, enterprises/establishments – Linking registers without unified codes possible, but much more difficult and time consuming • Comprehensive and reliable register system – Developed for administrative needs in society (social security, taxation etc) – Quality for administrative purposes generally good: w Register information exchanged between authorities w Quality important for individual’s rights and duties – Administrative quality necessary, but not sufficient for statistical quality • Cooperation among authorities – Statistics act: SN must be informed when administrative registers are changed – SN plays an active part when registers are initialised and established – SN supports register authorities w Technical support: Reporting quality problems w Political support: Call for resources needed 6

Development of register-based censuses Two main trends: 1. Extensive use of census data in establishing registers One way principle: Not statistical data for administrative purposes – Multiple legal bases necessary for collecting “census” data – 2. Step-by-step development Subject matter statistics tested and published in different areas – Register data introduced in censuses when quality is sufficient – Time from register established to data in census may vary – w w Labour market variables: 23 years Dwelling register: 6 -7 years? ( established in 2001) 7

Use of multiple sources • Data from one single register may not be sufficient for statistical purposes – – • Differences in definitions Limited coverage Errors Inadequate updating Using multiple sources in creation of new variables (derived variables) – Example: Employed persons • Prioritisation of information from different sources – Example: Current activity status • Editing – Correction in statistical registers – Report systematic errors to register-keepers w Not on an individual level 8

The Census 2001 data capture Labour market Social security GAB Income Education Persons: Registers CPR Families Couples Households Housing Census Traditional Dwellings Families and households: Registers and forms Housing and dwellings: Census forms Combined method in 2001: € 12 500 000 Registers only for 2011: € 1 250 000 (estimate) 9

Challenges using registers • Missing values • No or insufficient updating • The link between the resident persons and dwellings • For some topics: No administrative data exist – Human behavior and attitudes 10

Master registers and statistical registers • Three administrative master registers – Central Coordinating Register for Legal Entities (CCRE) – Ground Property, Address and Building Register (GAB) – Central Population Register (CPR) • Three statistical registers at Statistics Norway – Central Register of Establishments and Enterprises (CRE) – Register of Ground Parcel, Address and Building Register (SGAB) – Statistical Population Register (SPR) • The three registers are seen in connection to each other – Address as link key 11

Three versions of each register The owner of the register The base register The original Statistics Norway A true copy The statistical base register The statistical version – Statistical population register 12

What is the common identifier? Three base registers in Norway Businesses (legal entities) People Properties Addresses Buildings 13

Administration of statistical populations Original register Statistical version 14

Management of populations Online searching statistics Data collection Population files Data reception Prod. of statistics Input registration Sample survey admin. Reports SGAB SREE 15 SPR External administrative systems S Y S T E M S Data collection systems SAKSYS, DSF S T A T I S T I C A L IDUN, ALTINN, KOSTRA Do. T BRC NMA NIA Oth. Public administration Data suppliers 15

Assosiated registers to CCRE Brønnøysund Register Centre Register of Bankruptcies 1. 1. 07 Register of Business Enterprises Statistics Norway (SN) Central Register of Establishments and Enterprises (CRE) Directory of taxes VAT register Coporate Taxation Data Register Central Coordinating Register for Legal Entities (CCRE) Lottery Inspection Register of Foundations Norwegian Labour and Welfare Organisation Register of Employers 16

Aspects of quality in register based statistics • Relevance (for statistics) • Accuracy (of each record) • Timeliness (update of registers) and reference time • Comparability (over time) • Accessibility and clarity (documentation) • Administrative concepts • Coverage and completeness • Record matching ability • Compatibility (between file formats) • Confidentiality 17

Summary and conclusions • With the development of administrative sources in electronic form they should be used as much as possible for the production of official statistics • This development requires – Proper legal frameworks – Good understanding of benefits and possibilities by politicians and the public – Good understanding of the differences between administrative data and statistical data, concerning quality and usage – Good cooperation between administrative and statistical authorities 18