Philip CrossWayne Smith’s resignation as the chief statistician of Statistics Canada is the latest sign that the agency still has some maturing to do, writes Philip Cross.

By Philip Cross, Sept. 21, 2016

The resignation of Wayne Smith last Friday marks the second time in six years Canada’s chief statistician has quit in protest. Who would have thought that being the head of Statistics Canada, where I worked for 36 years, would become one of the most dangerous jobs in Ottawa?

It is all the more surprising because one of the very first things the Trudeau government did was to reinstate the long-form census and show support for Statistics Canada’s data gathering. This government will learn that gratitude is not a quality in ample supply in today’s civil service, which pursues its own self-interest with the passion it once reserved for serving the public’s.

Who would have thought that being the head of Statistics Canada, where I worked for 36 years, would become one of the most dangerous jobs in Ottawa?

The resignation of Munir Sheikh, the previous chief statistician, over the government’s handling of the census controversy was the culmination of months of discordant discussions between supporters of the census and the Harper government. The integrity of census data is an issue that the public can easily grasp, even if it turned very few votes (the Harper government won its first majority after Sheikh resigned in 2010).

It is hard to see Smith’s resignation provoking anything like the same response with the public or even Statcan’s most informed users. The issue revolves around the centralized provision from Shared Services Canada of informatics services that used to be done in-house at Statcan. Smith has argued publicly for months that this compromised Statcan’s independence. However, it is hard to believe Smith’s claim that Statcan’s “independence has never been more compromised.” Many things threaten its independence; the most existential would be publishing data or analysis simply to please the government of the day or to favour the opposition parties.

In my regular communications with people at Statcan this year, no one has mentioned Shared Services (although this might reflect my reputation for ignoring informatics issues). A litany of complaints written by Statcan’s managers early in 2016 seems clearly orchestrated; how else to explain that these complaints were all documented, just waiting for a friendly journalist to submit an access-to-information request? Meanwhile, the inundation of access-to-information requests about the census (Statcan had to hire two full-time employees just to handle all the requests) yielded nothing, because managers were instructed to not write down anything substantive.

Outside support for Smith’s position is going to be crucial. So far, this support appears to be quite limited, likely because this allegedly existential threat is hard to fathom. The government has demonstrated that it is quite sympathetic to Statistics Canada. Still, after listening to Smith’s appeals, it decided to side with the senior mandarins who maintain that Shared Services provides the best solution. Several members of the National Statistics Council (which provides nominal oversight to Statcan’s activities) expressed surprise about the resignation, implying Smith had not conveyed to them the gravity he attached to the issue. Resigning last Friday at noon was itself a bizarre tactic in currying media interest; Friday afternoon is traditionally when governments release news they want to bury, not the time to launch a crusade to galvanize public support.

The whole issue of informatics support from Shared Services is likely to quickly fade away. The new chief statistician (Anil Aurora, a capable manager with extensive experience at Statistics Canada) likely took the job on the understanding that he would work to find a solution with Shared Services, just as Smith himself took the job six years ago on the proviso that he implement the voluntary National Household Survey instead of the mandatory long-form census. With some creative thinking and an end to the public name-calling, solutions would seem possible. It helps to recall the uproar just last summer when Statcan published erroneous employment data due to its own computer-programming mistake; with such a fresh and embarrassing reminder of its own fallibility, one would think Statcan would be more tolerant of imperfections at Shared Services as new systems are worked out.

Independence for Statistics Canada is a two-way street. Statcan wants to be free from external interference with its internal operations. In return it has a responsibility to not abuse the taxpayer’s dollars and to be open and honest in its communications to justify the public’s trust. Opaque meeting minutes, orchestrating access-to-information requests into what was essentially self-serving propaganda and resigning over informatics support — painting it as equivalent to the high-profile issues surrounding the census — are all signs that suggest Statistics Canada has not matured enough in its communications and does not have a plan for the oversight necessary to justify full independence.

Philip Cross is the former chief economic analyst at Statistics Canada.