By Wayne Smith, Nov 2 2016
In 2011, Statistics Canada was put into an ‘iron cage of bureaucracy’, severely hampering its capacity to fulfill its mandate. The cage – Shared Services Canada (SSC) was created on August 4, 2011, to transform how the Government of Canada manages its information technology infrastructure. Statistics Canada was required to transfer to SSC all of its data centre, networking, telecommunications and e-mail infrastructure, the corresponding budget and staff directly connected to these functions, and a proportion of its internal services budget to support those functions. Statistics Canada had no meaningful opportunity to challenge this decision prior to its implementation. Statistics Canada was stripped of the authority to deliver informatics infrastructure and telecommunications services itself or to contract to third parties for these services. SSC is therefore an absolute monopoly with respect to the supply of these services to Statistics Canada. Shared Services Canada’s management, however, has no competence in matters of official statistics.
After several years of trying to make the new arrangement work, on August 3, 2016, I wrote to Prime Minister Justin Trudeau expressing my grave concern over the impact of Shared Services Canada on the independence of Statistics Canada, on Statistics Canada’s ability to protect the confidentiality of respondent information to the level required by the Statistics Act, and on the ability of Statistics Canada to operate effectively and efficiently. I noted that this was inconsistent with the current government’s public commitment to act so as to reinforce the independence of Statistics Canada. I advised the Prime Minister that, in the absence of any resolution to the issues raised, I would resign as Chief Statistician of Canada on September 17, 2016 in order to call public attention to my concerns.
This note explains the reasoning behind my letter to the Prime Minister and my determination to either resolve the issues or resign. It draws on previous writings during my tenure as Chief Statistician of Canada and my letter to the Prime Minister.
Independence of national statistical offices and the need to protect confidentiality of respondent information
First, some background on the notion of the need for independence of national statistical offices and their obligation to protect the confidentiality of respondent information collected for statistical purposes.
The notion of independence as applied to national statistical offices has been developed by international organizations notably, with Canada’s endorsement, by the United Nations (UN) and the Organisation for Economic Cooperation and Development (OECD). The UN addressed the issue in its Fundamental Principles of Official Statistics adopted by the General Assembly, in resolution 68/261 of 29 January 2014. In its second principle, the UN wrote:
To retain trust in official statistics, the statistical agencies need to decide according to strictly professional considerations, including scientific principles and professional ethics, on the methods and procedures for the collection, processing, storage and presentation of statistical data.
The OECD in its Recommendation of the Council on Good Statistical Practice, adopted by the Council of Ministers on November 23, 2015 and binding on Canada, recommended that members should:
Ensure professional independence of National Statistical Authorities. To this end, Adherents should ensure that the National Statistical Authorities:
- i) are professionally independent from other policy, regulatory or administrative departments and bodies, as well as from private sector operators, considering that professional independence of the producers of official statistics is essential for the production and the dissemination of objective statistics;
- ii) have the exclusive authority, as part of their professional independence, to decide on statistical methods and dissemination;
iii) are protected, through the inclusion of explicit provisions in statistics legislation, from political and other interference in developing, compiling and disseminating official statistics.
These provisions of the UN and OECD documents are broadly respected in the legislation and practice of developed countries with the unusual exception of Canada, which is otherwise considered a world leader in official statistics.
Both the UN and OECD have also addressed the importance of the confidentiality of information provided to the national statistical office by individuals, businesses and other organizations for statistical purposes.
The UN Fundamental Principles state:
Individual data collected by statistical agencies for statistical compilation, whether they refer to natural or legal persons, are to be strictly confidential and used exclusively for statistical purposes.
The OECD Recommendation echoes this viewpoint, recommending that members:
Protect the privacy of data providers (including individuals, households, enterprises, administrations, and all levels of government) and guarantee by law the confidentiality of the individual information provided and its use for statistical purposes only.
Canada’s Statistics Act is exemplary in its protection of the confidentiality of respondent data. The provisions of the Act limit access to confidential data to sworn employees of Statistics Canada. Section 6 requires employees to take an oath to respect the provisions of the Act. Section 17 makes confidential data holdings secret from anyone who is not an employee. Section 18 limits the use of confidential data to statistical purposes and explicitly prohibits access by courts and police. It provides inter alia that:
No person sworn under section 6 shall by an order of any court, tribunal or other body be required in any proceedings whatever to give oral testimony or to produce any return, document or record with respect to any information obtained in the course of administering this Act.
Finally, Section 34 makes disclosure of confidential information by an employee a criminal offence punishable by fines and imprisonment.
The role of informatics in official statistics and international precedent.
Second, let’s consider the importance of informatics and telecommunications infrastructure to a national statistical office, and international precedents for arrangements like Shared Services Canada.
Modern statistical offices are critically dependent on informatics to carry out their operations. Every phase of the statistical process is an application of informatics, from drawing samples, to data collection (internet questionnaires, computer assisted telephone interviewing, computer assisted personal interviewing), to data capture (optical character recognition), to coding and classification (automated coding), to processing, estimation and weighting, to confidentiality control, to analysis (modeling, projections, simulation), to dissemination (web sites). A statistical office is, in effect, an exercise in applied informatics that depends for its effectiveness on timely, reliable, affordable and effective supply of informatics services. Modern statistical offices are in constant evolution to increase their efficiency and effectiveness, as well as to respond to the evolving and emerging information needs of data users and stakeholders. Responsive informatics support is critical. Anyone who can control the supply of informatics services to a modern statistical office, effectively controls that office and its programs.
I am not aware of any developed country where the national statistical office has been stripped of all authority over its informatics and telecommunications infrastructure. There are two relevant instances I know of where a proposal was made to include the national statistical office in a centralized informatics initiative, the United Kingdom and Australia. In both, after consideration of the imperative of the independence of the national statistical office and of guarantees of the confidentiality of respondent data, the national statistical offices were appropriately exempted from inclusion in the centralization initiative.
Even in Canada, there are precedents that would support exclusion of Statistics Canada from mandatory participation in the Shared Services Canada initiative. On January 13, 2016, the Courts Administration Service and the Registrar of the Supreme Court of Canada, arguing their need for independence and confidentiality, were granted an exemption from mandatory use of Shared Services Canada by Order in Council. Agents of Parliament, such as the Auditor General, Official Languages Commissioner, and the Privacy Commissioner, were exempted from the inception of Shared Services Canada from mandatory use of that department’s services, presumably also for reasons of independence
Assessment of the impact of Shared Services Canada on Statistics Canada’s programs
In my letter to the Prime Minister, I noted that, despite the government’s public commitment to further protect the independence of Statistics Canada, that independence had never been more thoroughly compromised than it is at present as a result of Statistics Canada’s forced integration into the Shared Services Canada initiative. Shared Services Canada now has complete control of the critical informatics infrastructure supporting Statistics Canada. No significant change to Statistics Canada’s programs or methods can be made without Shared Services Canada’s concurrence, which is not guaranteed. The bizarre governance around Shared Services Canada creates no obligation on SSC to provide Statistics Canada with any level or extent of support, and creates no accountability to Statistics Canada for any adverse effects of its actions.
If Shared Services Canada refuses or fails to provide an essential service, Statistics Canada has no recourse of any kind and is prohibited from seeking other suppliers or provide the service to itself, even though it is amply capable of doing so. As a result, Shared Services Canada has an effective veto over any project, program or initiative of Statistics Canada that requires modifications to informatics infrastructure, and, in the world of official statistics, any significant change does. This is not independence.
I also observed to the Prime Minister that Canada’s Statistics Act provides strong protection for the confidentiality of information obtained by Statistics Canada from Canadians and their institutions for statistical purposes. Under the Act, this protection is not only from actors outside the federal government, but from all other elements of the federal government itself, including CSEC, CSIS and the RCMP. Under the provisions of the Act, Shared Services Canada should not be engaged in the management of confidential data unless the necessary staff have been ‘deemed’ to be employees of Statistics Canada by the Chief Statistician, and they should not be ‘deemed’ unless Statistics Canada has meaningful supervision of their activities while they are engaged in the management of these files. While initially I accommodated the involvement of Shared Services Canada employees in the hope that a reasonable governance would eventually be established—one that provided for meaningful supervision by Statistics Canada—at the time of my letter to the Prime Minister I had concluded this would never be the case. As a result, it was my conclusion that Shared Services Canada’s involvement in the management of confidential data obtained under the Statistics Act is inconsistent with the confidentiality provisions of Act and should cease.
My final observation to the Prime Minister was that the concerns raised were more than academic. Statistics Canada, at the months preceding my resignation, was being hobbled by the poor quality and high cost of the services provided by Shared Services Canada. To operate effectively and efficiently, Statistics Canada requires what it long had when it managed its own informatics infrastructure: seamless, effective, efficient, timely and affordable supply. Instead, even trivial decisions were finding Statistics Canada managers locked in endless discussions with phalanxes of Shared Services Canada managers, where reaching an agreement was no guarantee of actual performance of the agreed action.
Major transformational projects were being delayed by the failure of Shared Services Canada to provide the necessary infrastructure in useful time, pushing off implementation dates, idling project teams and forcing expensive work arounds. A planned redesign of Statistics Canada’s web site in time for dissemination of 2016 Census results was compromised due to SSC’s failure to supply the physical hardware in time, despite repeated senior meetings to confirm this would occur and reassurances that the necessary timeframes would be met. Other examples of significant delays of transformational projects include the Integrated Collection Operations System (intended to completely overhaul the way Statistics Canada manages and conducts its survey collection to increase effectiveness and reduce costs) and the Integrated Business Statistics Platform (designed to increase the coherence of business statistics while reducing processing costs).
Repeated demands were being received from Shared Services Canada for transfer of funds to pay for things for which funds had already been transferred or otherwise provided, or that were the clear responsibility of Shared Services Canada. Cost estimates were often disconnected from any reality, draining program budgets and risking cuts to statistical programs in order acquire necessary infrastructure. During my tenure, Shared Services Canada made unilateral decisions that put Statistics Canada’s programs and operation at high risk by, for example: failing to properly maintain the principal data centre on which Statistics Canada relies; cancelling service contracts for aging equipment, contracts needed to ensure rapid repair should key equipment fail; failing to maintain capacity in line with operational requirements; and glacial response times when Statistics Canada did agree to pay a second time for critical equipment.
Statistics Canada’s management would never historically, have tolerated the level of risk of a protracted failure of mission critical programs that is now being unilaterally imposed on it by Shared Services Canada. In the final months of my tenure, Statistics Canada was experiencing an increasing frequency and severity of incidents affecting its operations and the quality and accessibility of its data. This at a time when the government is asking for innovation and ever more data, and particularly more timely and granular data. The illogic of demanding more from an organization while undermining its ability to respond to those demands is obvious. I advised the Prime Minister that the ability of Statistics Canada to operate was at risk and that its stature as a world leading statistical office would eventually be compromised unless the situation changed.
Absent any meaningful response to these concerns and issues, I resigned as Chief Statistician of Canada on September 17, 2016. My purpose in doing so was to call public attention to a situation compromising the independence of Statistics Canada and its ability to protect the confidentiality of its information holdings to the level required by the Statistics Act and undermining its ability to operate efficiently and effectively in service to Canadians.
 This term was coined by Max Weber about 100 years ago. He was concerned about the rise of bureaucracy and the problems that this might create.