Industry Committee on April 11th, 2017

[The Chair, Liberal, Dan Ruimy]

We're going to get started, as we have quorum.

Welcome, everybody, to meeting 56 of the Standing Committee on Industry, Science and Technology. Pursuant to the order of reference of Tuesday, February 7, we are continuing our study on Bill C-36, An Act to amend the Statistics Act.

Today, from Our Right To Know, we have Margrit Eichler, the president, and from the Organisation For Economic Co-operation and Development, we have Paul Schreyer, the deputy director of the statistics directorate, who is with us by video conference all the way from France. Bonjour.

We also have, from the Statistical Society of Canada, Brian Allen, the past president and, as an individual, Jean-Guy Prévost, a professor in the political science department at Université du Québec à Montréal.

We're going to get right into it.

Margrit Eichler, you have up to 10 minutes.

[Dr. Margrit Eichler, President, Our Right to Know]

Thank you.

Thank you for inviting me to speak before you. Permit me to briefly introduce myself. I am a retired professor of sociology and equity studies. I taught at the Ontario Institute for Studies in Education at the University of Toronto until 2011. I am here in my capacity as president of Our Right to Know. We are a registered advocacy group with the mandate to advocate for the free conduct, communication, publication, and archiving of research. Our slogan is “Public Science for the Public Good”.

The major data gathering institution in Canada is Statistics Canada. The well-being of Statistics Canada is therefore close to our hearts. Although there are a number of issues that could be addressed, I will restrict myself to only one point: the relationship between Statistics Canada and Shared Services Canada. When we learned that the former chief statistician had resigned in protest over the lack of independence of Statistics Canada, we contacted him to learn more. While I have never met Mr. Smith face to face, there have been many written and oral exchanges. What we learned from these alarmed us. We then contacted a number of experts to compare their view of the situation with that of Mr. Smith's. We found no reason to doubt his integrity and veracity.

The minister, in his remarks during the debate in the second reading of Bill C-36, made it clear that high-quality data are needed to be able to make informed policy decisions. He makes a strong and convincing case that this requires independence of the national statistical service. If passed, the bill will increase the political independence of Statistics Canada. We strongly applaud the intent of the bill on this count. However, given such clearly stated intent, it is puzzling that there is no assurance of administrative independence.

Imagine that you were the chef for a huge gala dinner for hundreds of people. The contract has been signed. The overall framework has been agreed upon, the menu has been decided, the serving times have been set, the sous-chefs will be hired. Then, you find out that there is an unanticipated wrinkle: there is a housemaster who will determine which and how many pots you can use at what time, how many burners you may use at what time, and how many and which sous-chefs you may hire. In other words, you realize that you would be in a position of responsibility without the authority to make sure that the menu can be served as planned. At this point, you would probably tell your employer to cook the meal himself.

While Statistics Canada's job is infinitely more important and complicated than creating a gala dinner, however splendid it may be, the agency does find itself in a similar situation. Bill C-36 says this, and I will quote it in an abbreviated manner:

The Chief Statistician shall.... decide, based strictly on professional statistical standards that he or she considers appropriate, the methods and procedures for carrying out statistical programs.... [and] control the operations and staff of Statistics Canada.

However, according to Wayne Smith, Shared Services Canada has complete control of the critical informatics infrastructure supporting Statistics Canada. This amounts to an effective veto power on the part of Service Canada 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.

What are the consequences of this arrangement? They include delays of major transformational projects, for instance, the integrated collection operations system and the integrated business statistics platform; an unwieldy, user-unfriendly website; idling project teams; delays in delivering needed hardware; problems in maintaining aging equipment; cost estimates disconnected from reality; and unnecessary financial difficulties.

In other words, for a statistical agency, whose primary objective is the production and dissemination of data and information, Service Canada is an inefficient system. It does not allow Statistics Canada to operate at a peak level of performance. It wastes human and financial resources.

We consulted with my former classmate and distinguished CRC Chair Monica Boyd. She is an expert user of Statistics Canada data and has been seconded three times to Statistics Canada on a visiting senior fellow basis. She described to us three of the recent problems associated with the shift to the Service Canada platform: it has become very difficult to find information; how issues are arranged is not always logical; posting of information that normally was routine now appears to be erratic. Recently a major set of analytical papers produced at Statistics Canada could not be accessed for two weeks.

She considers the relationship between Service Canada and Statistics Canada as a cancer that is slowly affecting the entire system. We argue that this is probably due to the fact that Statistics Canada has a different structure and a different logic from the other departments that are serviced by Service Canada. Most departments deliver programs and services; Statistics Canada delivers data analyses.

We also want to mention that we're not aware of any other national statistical service in a developed nation that does not grant administrative independence to its statistical service. To pass the legislation without at the same time removing Statistics Canada from under the control of Service Canada creates a serious set of problems that cannot but hurt what we all want: a truly independent Statistics Canada.

We therefore strongly recommend that complete authority to run its own operations be returned to Statistics Canada in order to enable it to fulfill its duties as outlined in Bill C-36.

Thank you very much.

[The Chair, Liberal, Dan Ruimy]

We're going to move on to Mr. Schreyer from the Organisation for Economic Co-operation and Development.

You have up to 10 minutes, sir.

[Paul Schreyer, Deputy Director, Statistics Directorate, Organisation for Economic Co-operation and Development]

Good morning, Mr. Chair and committee members. Thank you for giving me the opportunity to appear before this committee. I will provide some comments on individual parts of the proposed amendments to the Statistics Act.

Let me start by saying that my overall impression of the changes is clear. The amendments constitute a significant step towards modern statistical law in close alignment with the “Recommendation of the OECD Council on Good Statistical Practice”. The council that adopted this recommendation in November 2015 consists of the representatives of our 35 member countries. It's the first OECD legal instrument on statistics, and all of our member countries, including Canada, have signed onto it. The implementation of its 12 recommendations and the associated list of good practices are regularly monitored. For me, it's a natural reference point when commenting on the proposed amendments to the Statistics Act.

The first comment I'd like to offer concerns our recommendation number 2, namely ensuring professional independence of national statistical authorities from other policy, regulatory or administrative departments and bodies, as well as from private sector operators.

I'd like to stress here that the professional independence that is sought by the recommendation relates to decisions about statistical methods and dissemination. In our view, this is perfectly consistent with some basic administrative dependence of the National Statistical Office on the executive branch in terms of defining a statistics program or more generally in terms of being responsive to the statistical needs of government and of society at large.

The amendments to the current Statistics Act explicitly confer more responsibilities on the chief statistician. They stipulate that he or she make decisions with regard to statistical methods and procedures as well as dissemination, and they add clarity concerning his or her terms of appointment. They also anchor in law the proposed statistics advisory council.

All of these measures reinforce the professional independence of the statistical authorities and are therefore very welcome in light of the OECD recommendation.

There is, however, one passage that seems to point in a different direction. The draft law in clause 4.1 states that:

Directives on any methods, procedures and operations may only be issued to the Chief Statistician by the Governor in Council, by order

What this says, which is true, is that the subsequent proposed sections impose a strict requirement to publish this order within 15 days, which represents a strong requirement for transparency. Nevertheless, interventions by the executive concerning methods, procedures, and operations remain possible by law. While there may be some sort of constitutional reasons for introducing this passage, it does appear from our perspective that this is at variance with the OECD recommendation.

Looking around the international scene, I also find that it's a somewhat unusual passage in the Statistics Law of OECD countries and appears in some respects stronger and broader in its potential application than are the provisions for ministerial intervention in statistical collections that apply in other Westminster democracies like Australia or New Zealand.

In the same vein, the unchanged provision in the law that indicates that the Governor in Council prescribes the questions to be asked in any census could be interpreted as incompatible with the professional independence of the chief statistician.

Another comment in this context relates to the selection of the chief statistician. The amendments to the law provide few indications about the process for selection, nor are any required qualifications for the position mentioned. The good statistical practices included in the OECD recommendations indicate both that the selection process should be described clearly by law and that the appointment should be based on professional competence only.

Such additions to the text may be worth considering unless there is another Canadian law that generally stipulates the need for demonstrated professional competence for this type of appointment. I'm not in a position to judge that.

Similarly, there is a bit of a question about the selection process for the members of the statistics advisory council. Again, anchoring the council, its principal functions, as well as specifics about the tenure or remuneration of its members in the Statistics Act is decidedly good practice, but there is no explicit reference to the professional qualities or profiles that council members should possess. As a key function of the council is to advise on the quality of the statistical system, explicitly stipulating the relevant professional competence would appear to be a useful thing.

Somewhat in passing, I notice that the size of the council has been set at 10. Now, while this appears to be near the lower end in international comparison, it should also be said that there is a wide variation among OECD countries here. For instance, the Irish National Statistics Board has eight members, the Swiss federal statistical commission has 11, Australia has 16, the U.K. has 14, France has about 40, and Germany has over 50. You have a wide variety of cases, and I guess much depends on the specific functions that these councils are intended to have, so there is no single good practice in this respect.

Another observation—and this is my final one—relates to our OECD recommendation number four that stipulates the protection of the privacy of data providers. The existing Canadian statistics law has very good provisions for the protection of privacy and so is fully compliant with the recommendation. My observation relates to the automatic disclosure of census records 92 years after a census is conducted. Now, we did not go through an in-depth analysis of statistical laws in OECD countries to find out about this point in other countries, but none of the laws we looked at briefly—with the exception of the United States—had a similar specific provision for general disclosure by default after a certain lapse of time. We don't consider the 92-year rule at variance with the OECD recommendation, but it is clear that there is an arbitrage between the long-term privacy protection for individuals and their families and other societal interests. Of course, that's an arbitrage decision that you as parliamentarians have to make.

I thank you for your attention and will be happy to respond to any questions you may have.

[The Chair, Liberal, Dan Ruimy]

Thank you very much.

We're going to move on to Mr. Allen from the Statistical Society of Canada. You have up to 10 minutes.

[Dr. Brian Allen, Past President, Statistical Society of Canada]

Good morning, everyone.

It is indeed a pleasure to be invited to appear before the Standing Committee on Industry, Science and Technology. Although the invitation to appear before this committee didn't specify, I view my invitation as representing the Statistical Society of Canada. I'm currently past president. The president, who might normally be invited to appear in my place, is employed by Statistics Canada, and so was recused.

The Statistical Society of Canada is Canada's only national scientific organization representing statisticians in academia, government, and industry. Its mission is to encourage the development and use of statistics and probability through research, education, and the development of public awareness of the value of statistical thinking. One of the bullets of our mission statement is directly relevant to today's hearings, namely, to ensure that decisions affecting Canadian society are based on appropriate data and valid statistical interpretation.

The Statistical Society of Canada has always had a close relationship with Statistics Canada. Indeed, its current president, Jack Gambino, is employed by Statistics Canada. Five additional past presidents have come from Statistics Canada: David Binder, 2005-06; Jane Gentleman, 1997-98; Geoffrey Hole, 1989-90; Martin Wilk, 1986-87; and Ivan Fellegi, 1981. Two of these, namely Wilk and Fellegi, also served as chief statisticians of Canada.

In addition, the society has benefited greatly from the involvement of many members from Statistics Canada. The society has six sections. Each focuses on an area of special interest to members. The survey methods section was one of the first two established, and it has always had dedicated support of members from Statistics Canada. This includes organizing invited paper sessions at the annual meetings and the organization and delivery of workshops on topics of interest to survey methodologists and graduate students.

The society believes that policy decisions should be based on evidence, and this usually involves data. Good data, whether from a survey, census, or other source, rely on both statistical design used to generate the data and on the choice of appropriate statistical methods to summarize the data. Administrative data—for example, tax data and health records—play an increasingly important role in official statistics. The use of appropriate statistical methods is important there as well. For example, the use of such data often involves record linkage, requiring both probability and statistics, and hence the importance of appropriate data and valid statistical interpretations.

The society participated in this committee's hearings, and lobbied extensively, regarding the cancellation of the mandatory long-form census in 2011-12. It also provided a letter of support of a private member's bill, to reinstate the mandatory long form, in early 2015. The society welcomes the opportunity to continue to provide its expert guidance on matters of statistical methods and procedures for the collection and summary of data.

In the brief time I've had to consult with colleagues, the views expressed on Bill C-36 have been unanimously positive. It's a step in the right direction, and it's making important changes to the Statistics Act.

I have a few recommendations, mostly related to how the Statistical Society of Canada can continue to make a positive impact on the collection and summary of official statistics in Canada.

The first recommendation is to have a formal search committee for chief statistician that would search widely, worldwide, for the best candidates. This short list of candidates would then be submitted for cabinet's consideration.

Second, consider consulting with the Statistical Society of Canada when seeking members to serve on such a search committee for the purposes of putting forward this short list of candidates.

Third, the Statistical Society of Canada is supportive of the establishment of the Canadian statistics advisory committee. It is recommended that it include members of the Statistical Society of Canada. This could be the president, but it might also be his or her designate.

I thank you and would be happy to answer questions.

[The Chair, Liberal, Dan Ruimy]

Thank you very much.

Now, we're going to move to Monsieur Prévost.

Mr. Prévost, you have the floor for 10 minutes.

[Jean-Guy Prévost, Professor, Political Science Department, Université du Québec à Montréal, As an Individual]

Thank you.

My name is Jean-Guy Prévost. I'm a professor in the Political Science Department at the Université du Québec à Montréal. Over the past 30 years, my research has focused on the census and on the functioning of statistical offices, not only from a Canadian perspective, but also from a comparative perspective.

The tabling of Bill C-36 arose largely from the 2010-11 census crisis, which resulted from the government's decision to remove the obligation to complete the long-form census and replace it with a national household survey, participation in which would be voluntary. The decision was consistent with the Statistics Act, but it raised a huge public outcry, in large part because the decision was made very quietly, meaning that the public wasn't informed until the matter was resolved.

The issue the requirement to respond and the issue of penalties for refusing to respond are, of course, political in nature, but the government's decision also had a technical dimension. We now know, with evidence, that despite the enormous resources provided by Statistics Canada, the quality and reliability of the results of a survey of a sample of volunteers is unsatisfactory.

The government's decision to impose its point of view led to the resignation of the chief statistician at the time, and that major event is one reason why the decision was considered a breach of the agency's professional independence.

We can legitimately wonder what the independence of a statistical agency is. This concept wasn't widespread in the 1980s, but it is becoming more common today. Of the 34 OECD member countries, 29 have adopted legislation that, in connection with Statistics Canada and its chief statistician, include explicit reference to the term “independence” or one of its correlates, such as “autonomy”, “objectivity”, “impartiality” and “neutrality”. The 1985 Statistics Act is one of five acts that do not contain such a reference.

The reason we have witnessed a movement in favour of an explicit reference to the concept of independence in statistical legislation is that independence and the appearance of independence are indispensable conditions for establishing the credibility of statistical data in a world where data should play an increasingly important role.

In addition to incorporating the idea of independence or one of its correlates into the legislation, OECD countries have resorted to other means to ensure their protection. For example, some have clarified the appointment and dismissal procedures of the chief statistician. Others have set up a body to advise the chief statistician and assess the quality of the work done by the office. Several other countries have opted for codes of best practice and quality assurance programs.

Lastly, we should keep in mind that independence, understood in the sense of protection against intervention by the political authorities in the professional or technical dimensions of statistical work, is never enough. A statistical office must also have clear mandates and authority as well as means to carry out its task. The more recent resignation of chief statistician Wayne Smith illustrates the complexity of these issues.

How does the Bill C-36 address concerns about the independence of Statistics Canada? The proposed amendments contain a number of welcome clarifications, but they also contain ambiguities. I will limit myself here to a few points.

The first point concerns the appointment of the chief statistician. The irremovability of the chief statistician for the duration of the term constitutes an appreciable protection, and there is an equivalent precision in several countries. However, nothing in the act confirms what is mentioned in the notes that accompany the bill, namely that this appointment follows an “open and transparent” selection process.

The second point concerns the professional independence of the chief statistician. Subsection 4(5) as amended reads: “The Chief Statistician shall ... (a) decide, based strictly on professional statistical standards that he or she considers appropriate, the methods and procedures for carrying out statistical programs ...”

This reference in the bill to what is generally understood as “independence” is the clearest to date—even though the term is not explicitly used. The list of specific items to which these professional statistical standards apply is consistent with what we have seen in other countries.

However, subsection 4.1(1) states: "Directives on any methods, procedures and operations may only be issued to the Chief Statistician by the Governor in Council, by order, on the recommendation of the Minister.”

It seems to me that we are going right back to what we saw in summer 2010: the possibility of intervention by the political authorities in the field that should come under “professional statistical standards” without prior public debate. If we absolutely want to keep the possibility for politicians to intervene on this issue, the only way, I think, is to allow public debate and to authorize the chief statistician of Statistics Canada to make his opinion publicly known. Otherwise, in the event of disagreement, the outcome will be the chief statistician's resignation, which ultimately results in a loss of public confidence. The fact that the section of the 1985 Statistics Act dealing specifically with census matters remains unchanged in this bill curiously leaves such a possibility open.

A somewhat different problem arises with respect to subsection 4.2(1), which refers to directives on statistical programs. It is legitimate for a government to want to obtain statistical information on areas previously ignored, or for budgetary or other reasons to end certain programs. The bill provides that the chief statistician may require that such directives “be made in writing and made public before the Chief Statistician acts on it.” It is difficult to understand why the chief statistician's ability to require the publication of such directives does not extend to those that would more directly affect what should be his or her area of concern, namely the methods and procedures.

The third point relates to the existence of a quality assessment and consultation body. The existence of such a body is also a way to strengthen and guarantee the independence of Statistics Canada. The National Statistics Council, which was established in 1985, has been an almost invisible structure: this doesn't mean that its members haven't played an advisory role to the chief statistician, but the council had no public profile—no website, two lines on the Statistics Canada website, no list of members anywhere, and so on.

The notes accompanying the bill provide the council with a considerable mandate, but as there are no resources planned to meet the required profile, there is no certainty that this new council will be able to do much more than the previous one.

Furthermore, Statistics Canada's independence is also based on means that exceed the legal framework, including, the body's adoption of a code of conduct comparable to the one that European countries have.

In conclusion, in a comparative study of all OECD countries, Canada appears to be one of the few countries where the independence of the statistical office is not legally or formally protected. While Statistics Canada has long enjoyed broad independence in practice, it was based on a gentlemen's agreement that ceased to exist in 2010.

In this regard, the bill brings two notable improvements: one to the chief statistician's irremovable character during the term of office and the other to recognition of his or her authority over professional standards, methods and procedures.

However, the bill should be clearer on these two points: with regard to the first, by including a reference to the selection process for appointing the chief statistician; in the case of the second, either by deleting section 4.1, thereby recognizing the exclusive jurisdiction of the chief statistician regarding methods and procedures, or by providing, as mentioned in section 4.2, that directives on methods and procedures “should be made in writing and made public before the Chief Statistician acts on it.”

At the present time, there is still a considerable gap between the explanations published on the department's site that reflect a considerable ambition and the bill itself. Basically, in contrast to the current legislation, this contains fairly modest changes to the independence of Statistics Canada.

Thank you.

[The Chair, Liberal, Dan Ruimy]

Merci beaucoup. We're going to go right into questioning.

Mr. Arya, you have seven minutes.

[Chandra Arya, Liberal, Nepean, ON]

Thank you, Mr. Chair.

Thanks to all of you for being here.

My first question is for you, Ms. Eichler. You mentioned problems with making the website user-friendly. These are all fixable problems. One of the most important benefits, in my view, of Stats Canada's use of Shared Services is to prevent cybersecurity threats. That is a real threat today. Maybe I missed it, but I noticed that you didn't mention anything about cybersecurity and how Statistics Canada, working with Shared Services, will make cybersecurity issues more preventable.

[Dr. Margrit Eichler]

That's an area outside of my area of competence, so I didn't comment on that.

[Chandra Arya, Liberal, Nepean, ON]

Thank you, Madam.

Mr. Schreyer, in your view, how does this bill compare to those of other countries such as New Zealand, the Netherlands, and the U.K.?

[Paul Schreyer]

I think it compares quite favourably now, in its revised form, with one or two elements that I mentioned, specifically, for example, this proposed section 4.1. One does not find that in the same form in, say, Australian or New Zealand law.

[Chandra Arya, Liberal, Nepean, ON]

This bill makes the census of population mandatory.

How do other countries do this? Having the Governor in Council with exclusive authority to determine the census questions, how does that compare with other countries?

[Paul Schreyer]

Well, it varies between countries.

Very often, you do find an explicit requirement to conduct a census; the requirement is laid down in the law. More generally, of course, a minister can issue the request to compile some form of information or statistics that he or she may need. What is not normally the case is that the minister would exert some judgment about how exactly this information should be compiled, and would interfere in terms of the methods that are being used.

You see in some OECD countries you have censuses that are conducted purely with administrative records. Denmark is—

[Chandra Arya, Liberal, Nepean, ON]

Okay, in fact, I was coming to that.

In the Netherlands, the last census I believe was done in 1971, where most of the data was collected from population registers. How does that work out?

[Paul Schreyer]

Well, you do have countries.... I mean, you mentioned the Netherlands, Denmark, and there are others, that do have a statistical system that allows them to make maximum use of administrative and other sources. They do not need the traditional ways of conducting censuses, but use the information they get from the records to have all the the information that is needed for society and political purposes.

[Chandra Arya, Liberal, Nepean, ON]

With your knowledge of the Canadian system, do you think that is possible here in Canada?

[Paul Schreyer]

I have no view on this. I don't know enough about the Canadian system.

[Chandra Arya, Liberal, Nepean, ON]

Okay.

[Paul Schreyer]

Typically, you do find these administrative registers, however, in rather small countries, with very specific legal environments that allow these crossing of data.

[Chandra Arya, Liberal, Nepean, ON]

You mentioned that the number of members on advisory councils varied from eight in Ireland to about 40 in France.

In addition to these advisory councils, do they have other committees that advise the chief statistician? For example, here in Canada, we have 13 advisory committees, plus around seven federal-provincial-territorial committees. Assuming that each committee has around 10 members, we have about 200 members advising the chief statistician. In relation to that, now we have this advisory council of an additional 10 members. Do other countries have these kinds of committees?

[Paul Schreyer]

Quite a few countries do have committees that operate at a more technical level. You have a committee under consumer price index, and another one on surveys or national accounts.

I do know that these exist in Canada as well, but their functions are typically somewhat different from the overall advisory functions that these statistical councils have. Indeed, you are right that the comparability may be limited. When we talk about the council in France, that includes, at least partially, some of these technical functions that you've been alluding to. For example, with Switzerland, I am part of the Swiss Federal Council. It has 10 members, and it works exclusively at this more strategic level.

[Chandra Arya, Liberal, Nepean, ON]

Okay, thank you.

Mr. Allen, thanks for coming here.

I'm glad to hear that most of your members are unanimously positive on this new bill. You mentioned that your first recommendation is to have a search committee. Now, with your knowledge of how the chief statisticians have been selected in the previous years, is there any deficiency in the current system? Were any of the selections made during the last 10, 20, 30 years—?

[Dr. Brian Allen]

I'm certainly not an expert on the history of the selection of the chief statisticians. I do know one positive example I'll cite, and that is Martin Wilk who was recruited in the 1980s.