Statistical Uses of
Administrative Data in
Canada
Presentation to OG6
Canberra, Australia
May 2011
1
Overview of presentation





2
What is administrative data?
Applications of administrative data
Sources
Advantages and disadvantages
Examples of use in the energy statistics program
at Statistics Canada
What is administrative data?
 Records collected and maintained by other
government agencies to meet legislative or
regulatory requirements:
• To monitor activities related to the production and
consumption of energy
• To enable regulatory or audit activities
• To assess outcomes of policies, programs, initiatives
• To register particular events (e.g. business
incorporations, licensing)
3
This differs from statistical data
 Statistical records:
• Are used to study dimensions, trends and
relationships of groups of persons or entities
• Maintain the confidentiality of individual identifiable
information
• Are used to produce aggregates, averages, measures
4
Applications of administrative data
 Direct collection of data
 Substitution for survey data (e.g. for nonresponse, for replacement of strata)
 Estimation, edits, imputation
 Creation and support of survey frames
 Data analysis, validation
 Improving timeliness
5
Sources of administrative data
 Other government departments and agencies
responsible for regulation, audit, evaluation
• Can be national or sub-national
 Can be made available through:
• A file of individual records
• Aggregated data
• A survey or census of administrative agencies
6
Advantages of using admin data




Reduction in the overall costs of collection
Reduction in burden placed on respondents
Greater availability of small area data
Possible reduction in errors due to census
coverage
 Opportunities to link data to other admin sources
 Advances in technology make it easier, cheaper
to manipulate and use large data files
 Useful in the development of survey frames
7
Disadvantages of using admin data
 Admin data are not primarily collected for
statistical uses, which could lead to :
• Inconsistent concepts, variables, definitions
• Lack of follow-up, quality control
• Possible breaks in series with changes to regulations,
legislation, policy
• Non-representative population
• Timeliness concerns
 Does the statistical agency have a legal right to
access? (Statistics Act)
• Public concerns over confidentiality, intrusiveness?
8
Examples from the energy
statistics program at Statistics
Canada
 Using administrative data ...
9
For direct collection
 Where survey data do not exist
 Eg. Deregulation of the energy market created
secondary distributors
• Data collected from natural gas distributors and
transporters on sales to residential, commercial and
industrial sectors
• A proxy for consumption data
 Eg. Import and export data from the National
Energy Board
• From admin files on permits
10
For direct collection
 As a supplement for existing data
 Eg. For monthly crude oil and natural gas
production, imports, exports, domestic deliveries
and inventories, data are collected through both
survey & admin data
• Producing provinces gather regulatory data
11
As a substitution for survey data
 Eg. Electric Utility Financial Report gathers an
annual breakdown of a series of balance sheet
and income statement items
• In 2008, negotiated with Ontario Energy Board to
gather aggregate admin data for 85 municipal
respondents
• Reduced costs, burden
12
For estimation, edits, imputation
 Where data are missing, incomplete,
inconsistent with historical data
 Eg. Missing fuel consumption values are
calculated using the change in fuel consumption
by industry
13
To support survey frames
 All energy surveys recently hooked up to the
Statistics Canada Business Register
 A good source of information to establish and
maintain survey frames
 Promotes harmonization across department
14
For data analysis and validation
 Energy statistics program uses other sources of
information for quality control, outlier detection,
confrontation
 Examples:
• Reports from industry associations
• Reports from other government departments
15
For improving timeliness
 Program is working to accelerate the collection
and reporting of key petroleum production data
• Using data from the Petroleum Registry of Alberta
(PRA) as a starting point, to eliminate duplicate
reporting
• PRA collects data for regulatory purposes
• The application may be extended to other producing
provinces
16
For the future...
 Other applications under investigation:
• Tax data, to replace financial reporting
• Public sector statistics, for data on electric utilities –
operating revenues and expenditures, sales of
electricity, using income statements and balance
sheets
 Currently assessing survey data vs admin
17
Conclusion
 Statistics Canada is relying more on admin data
 Driving factors: fiscal restraint, evolving
technology, respondent burden
 Success based on legislative authority +
ongoing collaboration
 Need to be aware of the constraints and
challenges of using admin data
18
For more information
Andy Kohut
Director, Manufacturing & Energy Division
Statistics Canada
11th Floor, Jean Talon Building, section B-8
Ottawa, Ontario CANADA
K1A 0T6
613-951-5858
Andy.Kohut@statcan.gc.ca
19