Transcript Statistics NZ`s experience in using Administrative Data in an

Statistics NZ’s experience in using
Administrative Data in an Integrated
Programme of Economic
Vince Galvin
General Manager
Strategy & Communications
Scope of Presentation
• New Zealand’s system of economic
measurement
• Measurement challenges and the role of
administrative data
• Examples of Statistics New Zealand’s
use of administrative data
The scope of User Need
The New Zealand System of Economic Statistics
Sustainable
Development
Indicators
Other Satellite Accounts
SEEA
System of National Accounts-Conceptual
Integration
National Accounts (incl BoP,GFS)
Business Indicator Statistics
Price Indexes
Business Structural and Performance Statistics
Business Register
Classifications/Definitions/Methodologies
Statistical
Surveys
Administrative Record Data
(e.g. tax, Customs)
businesses/households
Significant Issues
• Quality Issues
- “Tired” Designs
- Needed Integration not “designed in”
• Data Gaps
• Capturing structural economic changes effectively
• Growing demand from other users
– Sub populations, cross cutting topics
– Financial position
– Linking financial performance with business behaviour
• Manage Respondent Burden
– Reduction, Communication, Taking advantage of new
technology
What might the future look like
• Small Businesses monitored by administrative data
• Administrative data used to monitor populations, provide core
data across the economy for all units and provide essential size
measures for sample surveys.
• Sample surveys focus on complex data for large businesses
• Data integration studies to meet detailed analysis needs.
Key idea is that the administrative data goes from being
a useful data source to the population definer and major
source.
Reasons for Pursuing this idea in NZ
• Characteristics of Administrative data well
matched to population management
purposes if we can integrate multiple sources
• Tax data has value as a “measurement
foundation” across industries. Small
businesses are simple entities.
• We have had considerable success in
compiling integrated data sets.
The Potential Contribution of Administrative
Data
• Key challenge to manage information about business
population dynamics.
–
–
–
–
Measurement done consistently across industries
Size and Industry up to date
Trace the status of key assets
Sub population membership
• Tax data in conjunction with other sources
– Can trace dynamics of population but can’t provide detailed
data on large complex businesses
– Can provide core data across the economy
The Coverage of Administrative data
O
P
Q
Education
Health & Community Services
Cultural & Recreational Services
Personal & Other Services
B
C
D
Agriculture, Forestry & Fishing
Mining
Manufacturing
Electricity, Gas & Water Supply
Construction
Wholesale Trade
Retail Trade
Accommodation, Cafes & Restaraunts
Transport & Storage
Communication Services
Finance & Insurance
Property & Business Services
Government Administration & Defence
N
A
Coverage of SNZ Subannual Industry Collections
Industry (ANZSIC Division)
E
F
G
H
I
J
K
L
M
QMS
QBAS WQS RTS RTS
QEAS
RTS(pt)
QES QES QES QES QES QES QES QES QES QES QES QES QES QES QES
QES
GST GST GST GST GST GST GST GST GST GST GST GST GST GST GST GST
GST
EMS EMS EMS EMS EMS EMS EMS EMS EMS EMS EMS EMS EMS EMS EMS EMS EMS
Value beyond Survey Data
• Linked Employee/Employer database (LEED)
– Understanding dynamics of labour market
– Contribution to Statistical Infrastructure
(geography in business frame, BF structures)
• Longitudinal Business Frame (LBF)
– More detailed understanding of the Demography
of the business population
But of course……….
• The main challenge is identifying a map of the key
needs and their implications for the relationships
between data sources. Having a serious look at the
annual economic collection as a “hub”
– Pilot study of how to design links between annual financial
performance data and business behaviour data.
– Rationalising the quarterly collection strategy
– Annual Economic Survey relationship to quarterly
collections.
– Reconsideration of how we use the random number line to
minimise sample overlap
– Relationship between Commodity information and financial
information.
Example 1: The Business Frame
• Range of Improvements implemented
– Updated size measures, link to population metadata, use of
wide range of tax data as “triggers”
• Considerable benefits
– Reduction in birth timeliness lag, timeliness of size measure
updates, reduced direct surveying, improved population
analysis
• Process has enabled introducing new administrative
data
• The business frame becomes a centre of
“concordance management”
• Investing in Business Frame Improvements opens
the way for the widespread use of administrative data
Business Frame in SNZ
Business
Surveys
National
Accounts
• Wholesale/Retail Trade
• Manufacturing
• Production Surveys
• Annual Enterprise Survey
• Annual Frame Update Survey
•BAI
Balance of
Payments
BOP
Surveys
Labour Market
and Household
Economics
Population
Census
Quarterly
Employment
Survey,
LEED
Business Frame
Geographic Frame
Classifications and Standards
IRD
IRD
IRD
IRD
Example 2: The Use of Administrative Data
in Survey Design
•
Estimating economic variables from Tax data
– Projecting values in the Retail Trade Monthly Survey
– Modeling Quarterly Information
– Annual Financial data
•
•
•
Confronting high level aggregates
Compliance Cost Reductions
Improving Methods
– Assigning priorities in non-response follow up
– Improved imputation
•
Producing Regional Indicator information
The main question is now how the availability of data
and our information system improvements should
change the shape of our survey collection programme.
REDUCTION IN BUSINESS COMPLIANCE COSTS
DUE TO THE USE OF TAX DATA
SIZE
%
AFUS
45,000
34
MFUS
44,000
60
AES
9,600
95
20,000
50
QMS
600
36
RT
1,000
25
WT
360
20
(Sole traders and
partnerships)
AES
(Companies)
Example 3: The Agricultural Census
• Resumed after a gap of 7 years
• Considerable discussion as to how to construct a
frame
• Decision taken to use the same approach
– Standard approach helps put units into a slot across the
whole frame, using other sources to check coverage
– Better to take a frame that “over covers” than “under covers”
– A set of operational “scope and coverage” rules had to be
constructed to provide practical ways of bridging the gap
between concepts and limitations
To be Continued…………..
•
The extent to which the collection programme can be re-shaped is
being considered.
•
Information demands increase, compliance cost concerns still
significant
•
The increased use of administrative data in individual surveys has
provided a useful set of experiences in terms of understanding the uses
to which administrative data can be put in terms within our current
framework
•
The uses have highlighted that it is sensible to consider going further in
basing our collection strategy around administrative data. Attitude has
been surprisingly important.
•
The investment in the Business Frame and success with data
integration projects had put some of the infrastructure in place which
has created options.