eurostat results from the test egr with respect to inward and outward fats populations (2011)
TRANSCRIPT
EurostatEurostat
Results from the TEST
EGR with respect to Inward and Outward FATS populations (2011)
EurostatEurostat
Agenda
Objective of the TEST
Quality indicators on micro data
Structure of the test, 2 topics
Results on Inward and on Outward FATS
Some conclusions – next plans
EurostatEurostat
General objectives
Scope of the EuroGroups RegisterTo offer European statistical compilers coordinated
frame populations for statistics on globalisation
Vision for EGR Data Quality ManagementAn integrated process including statistical users to input the best available information into the EGR
The ultimate success of the EGR project is dependent on the quality of the frame
EurostatEurostat
Objectives of the TEST
Test the quality of the EGR 2011as a statistical frame population vs. Inward and Outward FATS 2011
• Establish cooperation between BR and FATS • Improve the capacity of BR staff vs. users’ needs• Support I and O FATS compilers to use EGR
• Topic 1 EGR 2011 / Inward FATS 2011• Topic 2 EGR 2011 / Outward FATS 2011
EurostatEurostat
Statistical frame quality indicators 1/2
• Across different sources (one is considered as benckmark)• E.g. EGR vs. Inward – Outward FATS populations
• Across different countries• E.g. Mirror exercises Inward vs. Outward
• Over time• EGR longitudinal analysis
Statistical frame - different dimensions:
EurostatEurostat
Statistical frame quality indicators 2/2
• Macro – data aggregated at NACE Division, by Size class, for Geographic areas, etc.
• Meso – clusters of data
• Micro – single unit (micro data linkage necessary)
Different level of data aggregation:
EurostatEurostat
Statistical frame quality terminology
• Completeness: a SBR is complete if it contains all units in the target population
• Coverage: SBR coverage defines the target population in term of size and activity (quality of coverage can be set differently according to users)
• Accuracy: a SBR is accurate if it correctly reflect reality (characteristics)
EurostatEurostat
Structure of the Test - InwardDefinition of the populations•EGR country specific Frame 2011: all resident units under foreign UCI•National IFATS population: population for IFATS used for survey, with values of UCI corrected during the survey (only not resident UCI) and used for the publication of the final statistics (ref. year 2011).
Methodology to compare the two populations•Micro level based (units merged by a common key)•Countries of UCI compared for the units in common
EurostatEurostat
Topic 1
I-Fats 2011
A
E
B
C
Z
(country-UCI correct) (country-UCI
wrong) D
(UCI =) (UCI ≠)
EGR Frame 2011 (country-UCI correct)
RESULTS
1 - Completeness of EGR = (B+C)/(B+C+D)
#DIV/0!
2 - Accuracy of UCI in EGR = (B)/(B+C)
#DIV/0!
3 - Coverage of EGR by country of UCI = F/G
SEE SHEET EMPL UCI
• Final IFATS population• Link resident units at micro level• COUNTRY of UCI, not exact UCI
EGR/I-FATS
EurostatEurostat
Definition of the populations to be compared•EGR country specific Frame 2011: all resident UCI •National Outward FATS population: resident UCIs used at the beginning of the survey and corrected after the survey
Methodology to compare the two populations•Micro level based (resident UCI merged by a common key)•Exact UCI compared
Structure of the Test - Outward
EurostatEurostat
Topic 2
EGR Frame 2011 O-Fats A D
B
Z
(UCI correct) (UCI wrong) C
E
(UCI =)
(UCI correct) (UCI wrong)
RESULTS
Indicator on the frame population of units (UCI) - % value
1 - Completeness of EGR UCIs = (B)/(B+C) #DIV/0!
Indicators on the countries of affiliates of the UCIs only for subset (B) - % value
2 - Completeness of countries of affiliates in extra EU27 = (F∩G)/[(F+G)-(F∩G)]
#DIV/0!
3 - Completeness of countries of affiliates in EU27 = (H∩I)/[(H+I)-(H∩I)]
#DIV/0!
• Initial OFATS population• Link UCI at micro level
• Some MS have Reporting Unitsdifferent from UCI > from REP to UCI and then link to EGR UCI
EGR/O-FATS
EurostatEurostat
Participating countries
• ESSnet countries: IT, NL, PT
• Countries with individual Grants: FI, NO, RO, SE
• Volunteer countries: LV, SI
Total: 9 countries
EurostatEurostat
Participating countries
• ESSnet countries: IT, NL
• Countries with individual Grants: FI, NO, RO, SE
• Volunteer countries: LV
Total: 7 countries
EurostatEurostat
• INWARD• Completeness measured at micro level LOW (≈ 40%)• Accuracy for the units linked HIGH (≈ 85%)• Coverage on employment VERY HIGH (>95%)
Problem of linkage (Set A vs D)
Units in EGR 2011 are all large
It doesn't mean that EGR covers ALL large units
(because we should measure employment of D)
But it is unlikely that very large units are missed
(EGR validation, profiling)
Conclusions
EurostatEurostat
Conclusions
• OUTWARD
• Completeness of UCI VERY LOW (≈ 20%)
• Completeness of countries IN EU VERY HIGH (≈ 99%)
• Completeness of countries OUT EU HIGH (≈ 70%)
Problem in the definition of the lists of UCIs
Methodological differences for choosing UCI
MORE DIFFICULT analysis, because we asked for the
same EXACT UCI
EurostatEurostat
What next?
• Continue: we expect improvements with EGR 2.0
• Add other indicators?• Inward: Empl.A/Empl.D• Outward: N. countries affiliates not in B
• Add metadata to the results?
• Publish the results of this test?