Chapter 3.5: Acquisition, use and management of administrative data and alternative data

Context

Administrative data are information collected by other organizations and departments, not by a national statistical office (NSO), for their own (usually non-statistical) purposes.

Using administrative data allows NSOs to:

  • improve the quality of statistical products (e.g., reduce sampling or non-sampling errors);
  • improve the relevance of statistical products (e.g., produce more detailed or more frequent estimates);
  • fill data gaps;
  • reduce the costs of statistical products; and
  • reduce response burden.

NSOs have been using administrative data for a long time, for example, in the compilation of National Accounts and for the recording of vital statistics such as births and deaths. In the context of modernization, some NSOs have begun increasing their use of administrative data for the production of statistical information.

In Canada, the Statistics Act of 1918 created a central statistical system, and paved the way for consolidation of the production of official statistics. A significant achievement in the area of vital statistics was noted in 1926: all provinces of Canada at the time were submitting their returns to the Dominion Bureau of Statistics. In the 1980s, Statistics Canada built the Business Register using Canada Revenue Agency income tax records and payroll deduction accounts data. In the 1990s, the agency started building an Address Register, which was first used as a coverage improvement tool for the 1991, 1996 and 2001census programs. Since 2006, the Address Register has been served increasingly as a frame for the census and household surveys, allowing, for example, questionnaires for the 2011 Census to be mailed out to 80% of private dwellings.

Statistics Canada has a long history of using administrative data in its statistical programs, when doing so leads to a better balance between relevance, quality, costs and respondent burden. In its long term vision, however, Statistics Canada plans to move to a model where administrative data become a foundational source of information for producing statistical information, using it whenever it is possible to replace questions asked to Canadians.

Statistics Canada defines administrative data broadly to include data from traditional sources; for example, data sets received from the Canada Revenue Agency, as well as data from alternative sources, such as data generated by electronic devices (e.g., satellites, sensors, scanners, mobile phones).

The Statistics ActEndnote 1 authorizes Statistics Canada to access all administrative data records that pertain to Canadian society or the Canadian economy held by organizations and departments, under

  • Section 13—Access to records held by federal, provincial or territorial departments; municipal offices; corporations, businesses and organizations; and
  • Articles 24 to 29—Specific access to income- and excise-tax records and to import, export and justice records.

However, although the agency can rely on legal authorities, the strategic approach of persuasion and a "win–win partnership" is the favoured one. This approach involves trying to find mutually beneficial arrangements to facilitate the exchange of information (e.g., value-added transformation of data) and the exchange of expertise. Each NSO must, therefore, serve as an influencer and play a proactive role in fostering changes that benefit both the statistical agency and suppliers of administrative data.

Strategies, mechanisms and tools

This section includes four important components that will help NSOs take greater advantage of administrative data in their information production process: (1) the definition of a corporate vision and strategic goals (2) the acquisition of administrative data (3) the use of administrative data, and (4) the management of administrative data.

1. Definition of a corporate vision and strategic goals

NSOs would benefit from identifying a corporate vision and strategic goals for the acquisition, use and the management of administrative data. To do so, each NSO must understand its own environment and emerging issues.

In Canada, as mentioned earlier, Statistics Canada has a long tradition of using administrative data in its statistical programs. The 2009 paper, A Long-Term Vision for Statistics CanadaEndnote 2,stated that "in the context of the kind of data we collect, we need to keep making progress on using administrative data as a substitute for survey data, where possible." To further develop this vision, a senior management working group, led by an assistant chief statistician, investigated the role of administrative data. In this context, the Committee identified and explored a wide range of statistical and strategic considerations regarding the use of administrative data, including the quality of administrative data, potential changes to legislation and government policy, privacy and public perception concerns, cost issues, and the implications on other government programs.

Following that work, a number of activities took place in Statistics Canada to further increase its use including the development of a policy governing the use of administrative data, creating an inventory of administrative data, building a research program, and establishing a clear governance structure were crucial steps to implementing a formal framework for the use of administrative data.

The need to enhance the use of administrative data has also been dictated by external factors. Among the most important was the creation of the Red Tape Reduction Commission by the Prime Minister of Canada, in 2011. The Commission was asked to identify irritants to businesses and to find ways to reduce business burden. Statistics Canada's 2012–2013 Report on Plans and Priorities stated that a key priority was the continued expansion of the use of administrative data to reduce response burden. Other factors, such as increasing demand for new data, fiscal constraints, and the need to find better ways to exploit existing data holdings, all contributed to the creation of the Administrative Data Secretariat (ADS), in September 2012, to develop and implement a corporate approach to increasing the use of administrative data. This mandate was passed to the Administrative Data Division further to the merging of ADS and the Tax Data Division, in April 2014.

To establish that corporate approach, the ADS carried out a review of international frameworks for the statistical use of administrative data. Drawing from an official publication of the United Nations Economic Commission for Europe,Endnote 3 the review compared Canada's legal, policy and organizational framework and statistical use of administrative data to those of five other countries (Ireland, United Kingdom, New Zealand, Australia and the United States). The study report included 25 recommendations for Statistics Canada to consider in developing its governance framework for the statistical use of administrative data. These recommendations were articulated around the following aspects: the legal and organizational framework, partnerships with data custodians, corporate tools, improved methods and processes, and communications tools. These recommendations served as the basis for a corporate strategy and corporate tools for Statistics Canada. The main elements of this strategy are presented in the following sections.

2. Acquisition of administrative data

2.1 Establishing and consolidating a governance framework

Legislation, policies and directives

As an initial step, each NSO has to analyze its specific legislation with respect to provisions that ensure the right to access, influence and use administrative data for statistical purposes. Policies, guidelines and procedures provide additional support. Legislation should reinforce the fact that the NSO should have access to and use administrative data, when possible.

In the case of Statistics Canada, these legal provisions include the Statistics Act, the Privacy Act, the Access to Information Act, and other acts that recognize the authority of the Statistics Act, such as provincial or federal legislation. These legal frameworks provide the opportunity to strengthen the agency's ability to access, influence and use administrative data.

Organizational and governance structure

The NSO should consider creating a centralized function for the acquisition of administrative data, at least those datasets broadly used in the NSO or complex to acquire. The responsibilities of this function and of other separate, but centralized, functions can include the following components:

  • Legal support service and information management;
  • Administrative data coordination for the management of corporate strategies, processes and tools; oversight of administrative data operations (capture, formatting/standardization, coding and validation of all incoming administrative data files); and
  • Creation and maintenance of base statistical registers (population, business and real estate), including the unique statistical identification of register units.

Whether the acquisition of administrative data is fully or partially centralized, the NSO must also identify and put in place the concept of a central administrative data custodian, i.e., the person / unit responsible for the following functions:

  • Liaising with the rest of the organization to understand the needs for the administrative data held by a given data provider;
  • Negotiating access to data provider's administrative data, holding delegated authority to sign and/or administer the agreement (including the respect of terms and conditions), and obtaining metadata associated with the administrative data;
  • Allowing others in the NSO access to the administrative data in accordance with the need-to-know principle; and
  • Receiving and handling questions from the data provider and other stakeholders as required.

In terms of governance structure, as the place of administrative data grows, the NSO should consider creating an Administrative Data Management Committee consisting of representatives from key areas of the NSO. This committee could be responsible for the following:

  • Overseeing the development and implementation of, and compliance with, corporate strategies relating to administrative data;
  • Determining the custodian of broadly used or complex-to-acquire administrative data files;
  • Making decisions regarding whether to pursue acquisitions of administrative data files that impose significant restrictions or costs on the NSO;
  • Reviewing administrative-data initiatives, and recommending changes to foster coherence and efficiency across the agency; and
  • Referring key decisions to the most senior management committee in the organization.

The operational arm of the Administrative Data Management Committee, including secretarial support, will normally be the centralized function described earlier.

2.2 Designing data acquisition agreements

As a best practice, NSOs should systematically document the terms and conditions related to the acquisition, management, use and disposal of administrative data.

This official documentation can take the form of the following:

  • Formal data acquisition agreements, including memoranda of understanding (same level of government) and agreements (different levels of government or private sector at no cost);
  • Contracts (with private sector when costs are involved);
  • Informal written communications (where permitted by legislation).

It is important to ensure that the document contain at least the following information:

  • The NSO's legal authority to obtain administrative data;
  • The NSO's intention to use the administrative data;
  • The NSO's legal obligation to protect the administrative data.

Documentation of agreements should be stored in a central repository.

2.3 Developing corporate tools

One of the objectives of NSOs is to develop corporate and common tools aligned with existing policies and directives. For Statistics Canada, this aligns as well with the Corporate Business Architecture; for more information, refer to Chapter 3.1: Corporate Business Architecture.

The following are the most important corporate tools for NSOs to develop and implement:

  • Administrative data inventory to register all incoming administrative data. The objective is to provide a central repository of information on administrative data holdings that can be viewed and used by employees through the organisation. For consistency and efficiency, maintenance and monitoring should reside with the administrative data centralized function.
  • Consultations to collect information on potential uses of administrative data files before an agreement is negotiated. The purpose is to ensure that consultations using a pre-defined template be conducted by the administrative data centralized function with support from program areas. This will ensure that NSO's needs for the data are well understood and that their use is maximized. This also provides an opportunity to avoid costly duplication of efforts on both the NSO side and the data provider side, which result from unclear needs.
  • Quality assessment tools to assist acquisition decisions. The administrative data custodian must conduct the assessment and store the results within the administrative data centralized function. This involves a two-step approach: first, using metadata and the information available from the organization to determine whether the administrative data can fulfill the needs of the NSO; and, second, providing a decision point as to the potential need to investigate further and to acquire administrative data or, at least, a sample or test version of the data. The process for assessing quality has to take into account the institutional environment of the data provider; that is, the organizational or institutional factors that may affect the data provider's capacity to supply quality administrative data over time from a legislative or reliability perspective. These assessments are to be consistently aligned with the various quality dimensions (relevance, accuracy, timeliness, accessibility, interpretability, coherence). Statistics Canada relies on its quality management framework. For more details, refer to Chapter 4.3: Management and access to metadata.

2.4 Managing strategic collaboration and partnerships

Managing strategic collaboration is key to the statistical use of administrative data. This approach can lead to better data and better methods and processes regarding use of the data.

The following actions can result in increased access to better administrative data:

  • Developing win–win partnerships with data providers, whereby access to the data by the NSO can result in
    • sharing non-confidential results on the quality of the administrative data with the data provider;
    • proposing to the data provider ways to improve the collection and processing of administrative data and the related documentation, which, in turn, serve the statistical use of the administrative data later on; and
    • returning the administrative data to the data provider once cleaned (with no confidential data added);
  • Working with other departments on government-wide policy changes that would facilitate access to federal administrative data;
  • Establishing bilateral committees with important data providers (at different levels) to ensure that there is no interruption in delivery and no unplanned changes to the data; and
  • Developing rules of engagement with private-sector organizations to secure stable access to their data sources by
    • promoting the benefits for them and their clients of reusing existing information rather than conducting surveys; and
    • considering tokens of appreciation, such as sharing with them non-confidential aggregates derived from their administrative data.

Finally, developing collaboration nationally or internationally with other NSOs, statistical organizations, the academic community, and the private industry can lead to better methods and processes for using administrative data—in particular, Big Data, which offer new, specific challenges.

Managing collaboration (internally and externally) and maintaining partnerships require effective two-way communications. This can be achieved by developing relevant and effective communications strategies and tools to explain the rationale behind the statistical use of administrative data. Communications channels may include website modules, internal committees, and consultation mechanisms that help inform and engage data providers and stakeholders.

3. Use of administrative data

Administrative data are collected by other organizations for their own purposes, which often are non-statistical purposes. Consequently, the usability of the administrative data will very much depend on the ability to align administrative concepts with statistical ones. This alignment may involve shaping the administrative data at the source, making compromises regarding statistical needs, or seeking innovative solutions.

The use of administrative data can be divided into two categories: direct use and indirect use.

  • Direct use refers to the immediate link between the administrative data and the statistical output. Examples of direct use include the direct computation of totals, means and percentiles as a stand-alone use of administrative data; substitution and supplementation for direct survey collection (with respect to certain variables, for all or part of sample units); and the production of analytical outputs.
  • Indirect use refers to instances where administrative data play only a supporting role in the creation of the statistical output. Examples of indirect use include the creation and maintenance of survey frames; the construction of sampling designs (stratification, sample allocation, sample selection); editing and imputation; estimation; and data validation or confrontation.

Whatever the use made of the administrative data, the NSO will have to document the statistical output and report on its quality (refer to Chapter 1.5: Managing quality).

4. Management of administrative data

NSOs must manage administrative data, while ensuring the following management practices are systematically and consistently considered and implemented:

  • Ensuring confidentiality and security – Agreements, legislation, policies and directives are fully respected. Re-using existing information (i.e., administrative data) is privacy-intrusive. This is why there are additional safeguards when it comes to linking administrative data, or disclosing to third parties, for example. For more details, refer to Chapter 4.6: Respecting privacy and protecting confidentiality.
  • Accessing and preserving data holdings – Limiting access to administrative data holdings to individuals in the NSO who have a demonstrated need-to-know is key. In addition, preserving the proper stewardship of information through the entire information life cycle requires effective information management strategies.
  • Disclosing information to a third party under strict conditions – Permitting access under strict conditions only, while underscoring the increased risk of residual disclosure. For more details, refer to Chapter 4.6: Respecting privacy and protecting confidentiality.
  • Communicating with the public and stakeholders — Informing, and engaging with, the public and stakeholders in a transparent and effective way, with a view to building a relationship of trust with data providers, privacy stakeholders and, ultimately, data users.

Key success factors

In addition to appropriate legislation supporting the access and the use of administrative data, collaborating and engaging with data-provider organizations and seeking win–win strategies are crucial to increasing the acquisition of administrative data and to expanding the use of administrative data. This partnership approach should be built and consolidated through inter- or pan-governmental policies and governance (by ensuring that all partners have input into setting priorities, and identifying and resolving issues). This is particularly useful when one is dealing with multiple partners (multi-jurisdictional).

Finally, because using administrative data is a priority for NSOs, clear directions must come from the top, and enablers—such as the ones described in this chapter—must be put in place.

Challenges

Some NSOs still struggle with developing their own governing legislation for the acquisition and use of administrative data. Establishing an enabling legal framework and building the appropriate organizational capacity are the first challenges facing NSOs.

In addition, managing relationships and partnerships requires continual focus, negotiation capacity, influence, and perseverance, since data providers do not have the same priorities as the NSO. Moreover, the data acquisition processes and agreements could be long and challenging endeavours.

Finally, infrastructure, expertise, resources, and dedicated funds must be in place to support the administrative data function and the corporate tools and processes available to programs areas—with a view to ensuring consistency and efficiency.

Looking ahead

For Statistics Canada, the way forward consists in supporting ,through a dedicated strategy, a key role for administrative data in the way NSOs produce statistics in the future,. In adhering to this approach, the agency aims to promote a whole-of-government approach to collecting information once, and to using it multiple times.

Endnotes:

Endnote 1

Government of Canada, 2005.

Return to endnote 1 referrer

Endnote 2

Sheik, M.A., 2009.

Return to endnote 2 referrer

Endnote 3

United Nations Economic Commission for Europe, 2011.

Return to endnote 3 referrer

Bibliography

Government of Canada (2005). Statistics Act. L.R.C 1985, c. S-19. Amended by 1988, c. 65, s.146; 1990, c. 45, s. 54; 1992, c. 1, ss. 130, 131; 2005, c. 31; 2005, c. 38. Consulted on the 11th of March 2016 and retrieved from http://laws-lois.justice.gc.ca/eng/acts/S-19/FullText.html

Sheikh, M.A (2009). A Long-Term Vision for Statistics Canada. Statistics Canada internal document. March.

United Nations Economic Commission for Europe (2011). Using Administrative and Secondary Sources for Official Statistics: A Handbook of Principles and Practices. United Nations: New York and Geneva. Consulted on the 11th of March 2016 and retrieved from https://www.unece.org/fileadmin/DAM/stats/publications/Using_Administrative_Sources_Final_for_web.pdf

Report a problem on this page

Is something not working? Is there information outdated? Can't find what you're looking for?

Please contact us and let us know how we can help you.

Privacy notice

Date modified: