How often do we hear or read the phrase “Statistics Canada reported today that…”? As the publisher of official statistics for Canada, we rely on Stats Can for valid and reliable facts. Surely this national resource is one antidote for the increasing flim-flam that marks current political and social discourse.
By any measure, Statistics Canada is one of the world’s premier statistical agencies. But before we conclude all is well with our system of official statistics, consider three challenges that are generally faced in the creation and publication of such information: interpretability, innovation and independence. How does Statistics Canada score on these quality dimensions?
To start with interpretability, consider the recent Labour Force Survey (LFS), which revealed a dramatic job loss for Canada, driven largely by employment decline in Ontario. This prompted the expected flurry of contradictory comments. Some pundits moved quickly to caution against making too much of a single month of this “volatile” measure. Since the losses were concentrated in Ontario, others were quick to finger the recent increase in the minimum wage as a potential culprit.
The issue of volatility does not result from defects in the survey but reflects a dynamic free-market economy. This does not render the LFS less useful, as I show here. When duelling commentators issue divergent narratives on a “fact,” what is the intelligent reader to think?
Started at the conclusion of the Second World War to provide policy guidance for the labour market integration of the returning armed forces, the LFS is conducted monthly and has a sample size of 50,000. It parses Canada’s labour force into three mutually exclusive categories — employed, unemployed and those not in the labour force. This seems simple, but complexities abound that make the survey a fertile ground for myth and misunderstanding.
One of the important services Statistics Canada offers is detailed technical documentation and metadata that accompany all its data releases. It would be helpful to public discussion if more pundits reviewed this material before expressing opinions. In my view, because of this documentation, Stats Can deserves high marks for interpretability and user support.
The technical documentation that accompanies all Statistics Canada’s data releases deserve high marks for interpretability and user support and more pundits should review it before expressing opinions.
On the second dimension, innovation, official statistics can innovate in two ways. First, data collection and analysis methods improve. An example of this is less reliance on surveys, which are costly, and more use of administrative data. Accessing tax files to replace direct questions about income on the census both reduces costs and increases data reliability. Perhaps a survey-based census will soon become obsolete.
Second, as society evolves, so does the need to find new measures to reflect those changes. “Inequality accounting” reflects a growing concern over how we measure economic well-being. While official statistics on poverty and inequality appear to be showing improvement, other evidence suggests that large segments of Canadian society are excluded from a growing economy. For example, the poverty rate as measured by Statistics Canada’s low-income cut-offs has consistently declined for the past decade. Also, the Gini coefficient for income, a common measure for inequality, which has declined recently, possibly shows Canada is becoming less unfair economically. If these indicators reflect reality, then where is the urgency for remedies such as the basic income and a higher minimum wage?
One way to deal with this discrepancy is to add more “texture” to inequality measures by apportioning changes in gross domestic product to different tranches of income. Paul Krugman offers a concise rationale for such a change.
Innovation is a good thing, but there is a constant tension between maintaining traditional topics and diverting resources for developing new measures. Recently when examining a thesis in agricultural economics, I remarked that the information base for the study seemed thin. My colleagues ascribed this to general fiscal restraint, but the more likely explanation is that Statistics Canada is grappling with greater data demands with limited resources. In other words, to innovate with new measures, Statistics Canada is cutting back on collecting data in more traditional areas. Wayne Smith, the former chief statistician, has expressed concern that Statistics Canada is not introducing the new data that are needed to properly reflect Canadian society.
Government determines the resources made available for official statistics, and this raises the third and most crucial quality theme: independence. On this dimension, I fear Statistics Canada is doing less well.
Statistics Canada has passed through a period when two chief statisticians have resigned, claiming some form of political interference. Appointed in 2008, Munir Sheikh resigned in 2010 to protest the Harper government’s cancellation of the long-form census. Wayne Smith, Sheikh’s successor, resigned in 2016 over a dispute where the government decided that, in the name of efficiency, Statistics Canada had to use IT systems from Shared Services Canada. Smith claimed that this would compromise the confidentiality of key data, notably information on the economy such as the LFS, which could influence money and stock markets if leaked prematurely.
The Liberal government reacted to Smith’s resignation with an amendment to the Statistics Act intended to increase the independence of Statistics Canada. Passed in December 2017, the key changes revised the language around the independence of the chief statistician in determining appropriate methods, formally recognized the role of a statistics advisory council, permitted the release of census records 92 years after data collection, and eliminated jail as a penalty for refusing to participate in mandatory surveys.
As Wayne Smith observed, the revised act still fails to ensure that the selection of the chief statistician is based on merit. It is possible for government to make a patronage appointment, although opprobrium would surely ensue. He also notes that future governments can still meddle with the form and content of the census. Finally, Smith notes that it is still required to use Shared Services Canada, which brings together “mandated email, data centre and network services to partner organizations in a consolidated and standardized manner to support the delivery of Government of Canada programs and services.”
Smith raises important points, but these are not enough. The government is reportedly open to pondering additional changes. Here are three proposals that will increase the independence of Statistics Canada.
- First, it needs to have the same status as the Office of the Auditor General and be designated as a Crown agency that reports to Parliament. Currently, the chief statistician reports to the minister of innovation, science and economic development.
- Second, the Statistics Act must specify merit as a basis for appointing members of the statistics advisory council, as it does for appointing the chief statistician. And, in addition to specifying that members of the advisory council should understand how to develop data that reflect our society, the Act needs to vest appointments to this council not with the government but with institutions likely have suitable candidates. Examples include the Royal Society, the Tri-Council agencies and the Council of the Federation.
- Third, since government can constrain the collection and publication of official statistics simply by reducing Statistics Canada’s budget, a formula-based approach to budgetary allocation would limit such discretion. For 2018-19, a noncensus year, Statistics Canada has a budget of $436 million, or about 0.16 percent of total federal spending. I propose to designate this as the base for setting future budgets for collecting official statistics. Adjustments to the formula would be possible to account for census years, when data collection costs understandably balloon. Annual adjustments could be linked to population increases to serve as a rough approximation for changes in Canadian society. While this will be challenging for any government to accept since no cabinet likes limits to its freedom in setting spending, it is essential to ensuring that Stats Can has true independence.
Statistics Canada has a justified reputation as one of the foremost official statistics agencies. On the three quality dimensions of interpretability, innovation and independence, it performs well on the first, faces growing challenges on the second and has considerable room for improvement on the third. These proposed changes to the Statistics Act will cement the independence of Statistics Canada and the chief statistician, and thus restore Canada’s commitment to truth telling.
Do you have something to say about the article you just read? Be part of the Policy Options discussion, and send in your own submission. Here is a link on how to do it. | Souhaitez-vous réagir à cet article ? Joignez-vous aux débats d’Options politiques et soumettez-nous votre texte en suivant ces directives.