Statscan’s Slow Demise

Philip Cross has become the latest causality in a series of losses to Statscan, the nation’s statistical agency. For Cross, the agency’s ex-chief economist, internal debates finally reached the point of no return. With multiple disputes over its storied history – most recently over the long-form census – Statscan’s stature and role in federal affairs has diminished. While many decry the quietly disappearing agency, we may want to ask whether its disappearance would mean anything to the average Canadian.

Tavia Grant, writing for the Globe and Mail, laments the loss of its chief economist:

“The departure of Mr. Cross, known for his blunt style, is significant as few others can sift through such volumes of data and make sense of economic cycles and what is driving them, said Mr. Sheikh, who created the role of chief economic analyst for Mr. Cross in 2008 – at a time when the economy was souring, and the public was hungry for insights into the turmoil.”

We would be remiss to think that it was Mr. Cross “sifting” through heaps of data that has kept Canada on a relatively smooth path over the past decades. The reason for our hesitation should become clear when we think two factors: the amount of data under inspection, and its relevance.

In his now classic 1945 article, “The Use of Knowledge in Society”, economist Friedrich Hayek pointed out the error in thinking that centralized data could be used for guiding an economy. He gave two reasons. First was the assumption that any central agency could ever know all of the knowledge that the market is generating. The market is a collaboration of decentralized knowledge – any individual only knows a small fraction of the total collective amount. The second reason was that this structure of knowledge was essential for its efficient dispersion throughout society. Coordination is efficiently undertaken on the market via distinct individuals producing, sharing and using their small pieces of knowledge. In this way, the data of the economy are created and dispersed among the relevant and disparate actors.

Statscan is by definition a centralized agency that collates disparate information so that it can be used. The limits on what the agency can accomplish hamper its role. Knowledge flows through the economy so quickly that its collection in a central agency is a dated endeavor. This is especially true of the census, which takes place only every 5 years. The market is a continual process of data generation and dispersion. The census is dated before it is even finished.

Much of the data that actually matters to entrepreneurs is not in a form that is easily quantified or transmitted. As Jesus Huerta de Soto outlines in his book, “Socialism, Economic Calculation and Entrepreneurship”, the data that matter most in the economy are mostly of a tacit nature. They are unable to be clearly articulated and transferred. Ask an entrepreneur why he think sales will be down this year. There are myriad plausible reasons why he has such feelings, but more often than not it boils down to a “hunch”, a “feeling”, or some foresight. These are the metrics that Statscan cannot collect.

What is the relevance of the data that the agency collects? Another one of Hayek’s great contributions was that aggregates and averages shield us from the important pieces of market data. In his debates with John Maynard Keyenes in the 1930s, Hayek stressed that Keynes missed the important facets of the business cycle by focusing on aggregates like gross domestic product, or the unemployment rate. The unemployment rate might tell us that today about 7.5% of Canadians in the labour force are without work. Statscan might tell us that this is an essential figure to know, so that policy makers can shed light on why we are at that level, or presumably even alleviate it.

But what does that 7.5% figure really tell us? For the individual, precious little. No one Canadian is “only” 92.5% employed. The individual is either employed or not. He is in this situation because of very specific attributes – his skills, productivity, education, place of residence, etc.  The business owner does not care if 7.5% cannot find jobs – he cares that he can find the right one for the job at hand. The unemployed don’t care too much of the figure either. They are concerned that they find a job, not that some mass of their peers are in the same boat as they are. Statscan recently reported that 250,000 jobs are vacant across the country. This information might make for an interesting headline, but for the unemployed it does nothing to help them get a job. The specific information of where, what skills are needed, and what type of work it is are all lost in the aggregate figure the agency reports.

In one storied tale, Sir John Cowperthwaite, the Financial Secretary of Hong Kong, refused to collect economic statistics of any type. Hong Kong was one of the great post-war success stories. Its economy witnessed growth rates envied by other countries. These growth rates were not driven by government policy (unless a hands-off approach qualifies as a policy). Its growth was driven by entrepreneurs making use of their specific knowledge to produce goods consumers valued. No top-down policy was necessary to achieve high rates of growth. The bottom-up approach managed to efficiently and effectively coordinate all the necessary pieces of information.

Where does this leave Statscan, or the departure of Mr. Cross from its ranks? Was the public really so “hungry” for its data and his analysis, as Ms. Grant reckons? Besides seeing the unemployment rate paraded around by politicians during election season, or debated on the floor of Parliament, for the individual it meant relatively little. He was concerned with his prospects when he opened the help wanted pages, and this was not information that Statscan provided.

Canada functioned as a country for over 100 years before the creation of Statscan in 1971. It flourished and achieved envious growth rates and prosperity, all without a central data collection agency. Since its creation, Statscan has “guided” Canadians through the muddle-through 1970s and early 1980s, the sluggish early 1990s, and the recent malaise. I can’t think of too many that would attribute the country’s boom years to Statscan’s services, or to Mr. Cross’ brilliant analyses. Which leaves us with the question: if the whole agency went away, would anybody notice?

Tags: , ,

2 Responses to “Statscan’s Slow Demise”

  1. David Howden says:

    Ohhh Henry:

    thanks for the comment. I guess there are two reasons one could look to for StatsCan's existence. I do think that by the government's own reckoning, if it is to allocate resources at the federal level it requires information about how to do so. In this case it makes perfect sense to have a statistics department to organize the necessary information (whether people want to provide, as you note, is another question). The question in that case is whether such a centralized information agency could ever provide the necessary information — which most definitely is a knowledge problem.

  2. Ohhh Henry says:

    The purpose of StatsCan is not to help entrepreneurs or to improve the standard of living for the Canadian public, it's a political tool created in order to assist politicians in finding and exploiting voters. Why else would they be so obsessed with racial statistics and so-called language rights? Not to mention personal incomes, the number and age of children and the quality of housing. It's a matter of finding a group of allegedly oppressed or neglected voters who are sufficiently concentrated in one or more ridings, so that by funding them with pork, or by whipping them up into a state of anger, their votes may be harvested. If these allegedly aggrieved voters are not sufficiently concentrated in any one riding, then the StatsCan data are used by politicians to re-draw the riding boundaries so that they ARE sufficiently concentrated.

    Another useful way for politicians to employ StatsCan data was shown by the recent robo-calling scandal where certain ridings with known voting tendencies were targeted for disruption – screwing up voters by lying to them about the location of their polling station. Anyone who considers the way that StatsCan data is collected and used, and who then considers how crooked elections are, would have to be an idiot to willingly participate in either a federal census or election.

    It is wrong to present the failure of StatsCan (from the public's point of view) as an example of the "knowledge problem". It's true that no large entity could ever collect, organize and publish the kind of information that entrepreneurs need. But that is completely irrelevant. The department was not created to assist entrepreneurs or to assist the public. The axioms of human action teach us that the politicians and bureaucrats who created StatsCan were acting purposefully for their OWN benefit, not for the benefit of anyone else. How could they do otherwise? Are they not human? The reason for StatsCan's failure to benefit the public is that the public is not a willing, voluntary participant in the creation or funding of the department, but is instead forced to support it under threat of violence, enslavement in prison and confiscation of property.

Leave a Reply

You must be logged in to post a comment.