Skip to content
How polyamorous households split finances

Three incomes, one mortgage: How polyamorous households pay

Northern Ontario breweries are going bust

​Northern Ontario's brewery boom is going bust

Search

Latest Stories

​Inside the Statistics Canada multiverse

Frequent revisions to official stats are confusing economists. Funding cuts could make matters worse.

A floating hand erases a line labeled 'recession' on a sheet of graph paper.

StatCan declared, then reversed, a recession. What's going on?

The Margin Staff


The dark art of statistics is not for the faint of heart. Its practitioners must strike a careful balance between the forces of accuracy and timeliness, appeasing the ravenous hordes demanding updates on the hour, while negotiating accords between conflicting figures and conjuring up estimates to fill troublesome gaps. Every month, the brains at Statistics Canada perform this feat to release GDP updates at the end of every month, each covering the month before. Businesses and economists trawl through them to judge the market’s temper, but they must do so with caution. With only 30 days to get the data together, StatCan can only do so much, and these monthly releases tend to be small on sample sizes and big on guesstimates.

That helps to explain why, this past September, an apparent miracle occurred. Overnight, Canada gained nearly 300,000 people, as though Thanos had snapped his gloved hand. It wasn’t even the only numerical apparition to bless the country that year: over the summer, a minor recession in the first quarter was retroactively reversed, spooking economists across the country. The spring, too, was transmogrified from a slump period into a solid season for growth, surprising even the high priests at the Bank of Canada, who had foreseen much grimmer tidings.

In our modern world, there is only one class of beings powerful enough to change the very image of reality.

They’re called statisticians. And this year, they’ve been busy.

Statistics have obtained reverent status in modern society: any undergrad knows that dropping a few hard numbers can transform a loosely slapped-together paper into a solid B+. But behind their veneer of permanence and authority, stats are actually very fluid things, and those from StatCan are no exception.

Take those monthly GDP numbers. Thirty days after its initial release for a given month, StatsCan puts out a second, more certain analysis, using more comprehensive methods. These are followed by reviews for the whole quarter, which have a bit more data to work with, but still come with drawbacks. Since they’re released one month after the end of the quarter – say, January to March – the data for the quarter’s third month (e.g., March) will only be some 30 days old. That leaves it open to all the flaws that come with flash reporting, including that some pertinent data may not be included yet.

These lags mean that StatCan is constantly playing leapfrog with the truth. In an attempt to keep up, they’re fond of correcting the record for previous months and quarters as time passes and they get new information.

These changes proceed on a rigorous schedule. Take the first quarter of 2023. Sixty days after it ended, in May 2023, it received its first quarterly estimate. Then, at the end of Q2, it got a second look. With Q3 came a grand revision of the previous three years, which included our quarter in question. Q4 then saw a revision for each quarter that year, ours included, and the same was done in Q1 of the next year, 2024. And that’s not all: our quarter was revised three more times in the trio of years following. Ultimately this means that any one quarter has, in StatCan parlance, eight “vintages,” with some, like dusty old bottles from the cellar, coming along four full years later.

Any one quarter has, in StatCan parlance, eight “vintages,” with some, like dusty old bottles from the cellar, coming along four full years later.

StatCan's website insists that "Revisions are not a sign of initial error; they reflect a commitment to timeliness, precision, transparency, and producing the best possible evidence." And yet, these fluctuations can create real headaches for top bankers. The Bank of Canada has some of the best economists in the country, but even they have found themselves shocked by StatCan’s revisions. In December 2025, the agency declared that GDP growth for the past three years had, in fact, been a whole percentage point higher than previously reported – enough to transform three years of declining growth into a mere plateau.

That change was the single-largest revision StatCan had made to its numbers in a decade, and it was a reality-bending one indeed. Up until then, the Bank of Canada had been labouring under the belief that GDP really was sliding, and they even cut interest rates to help combat the problem. As it turns out, those actions may not have been necessary – though they did prove helpful when the Canada–US trade war kicked off in 2025.

“Without good data, we’re all lost.”

There are signs of dwindling patience with these revisions to history. After StatCan published its recession finding earlier this year – but before it retracted it – two economists at Desjardins published a report bemoaning the “data quality crisis” afflicting the country. The unexpected finding, they wrote, “has led to a lot of soul searching among economists. How did so many forecasters get it so wrong?”

Their answer, which turned out to be correct, was that “monthly GDP is no longer a reliable indicator.” StatCan’s revisions have been getting “larger and more unpredictable” as the years go on, they found, which reflects the agency’s increasing penchant to publish preliminary findings and fix the numbers later. Its quarterly reports are published before real GDP data for the same period comes out, and in the absence of that data, the agency has to rely on models to estimate the truth. Any errors in the data then cascade through the rest of the economy, leading people astray all over. As the report puts it, “Without good data, we’re all lost.”

In a statement, StatCan defended their methods. "Revisions are required as the completeness of data sources increase and allow us to provide more up to date time series, while ensuring the continued quality of Statistics Canada data," said Carter Mann, the agency's chief of media relations. Their GDP figures for Q1 2026, Mann said, were revised upwards thanks to new numbers showing greater exports of non-metallic minerals and energy products. And the "larger than normal revision" to population numbers was due to new estimates of non-permanent residents and more administrative data. Despite the extra 300,000 people, Mann pointed out that their findings still lined up with a general trend: "Canada's population continues to slow down sharply."

In StatCan's defense, Canada is actually faring better in the reality distortion department than several other countries. Following the pandemic, public trust in the government fell to new lows, making it harder for official surveyors to pry data out of people. Canadians, by contrast, remain relatively trustful of the feds. It’s also not all StatCan’s fault, either: when it comes to international dealings like trade, the agency has to rely on datasets from other countries, and that leaves it vulnerable to any flaws in the foreign numbers it receives. The US has been especially unreliable lately: not only has the trade war disrupted information flows between the two countries, but nearly half of the data for the US CPI is inferred from models rather than actually measured. Those models may not get it right, but we have no choice but to rely on their results.

Fewer statisticians means fewer people gathering and crunching numbers, and without them, less certainty in the world around us.

StatCan itself is working under less than ideal conditions. We may scoff at Elon Musk's ill-fated Department of Government Efficiency (DOGE) or Javier Milei’s chainsaw-wielding antics, but the Canadian public service is also being forced to downsize. The Carney government expects StatCan alone to find $338.9 million in cumulative savings over the next four fiscal years, and then scrounge up another $100 million every year thereafter. Already the agency has cut its headcount for the coming years by 6%, and will have no choice but to do more with less.

"Statistics Canada has made targeted adjustments to certain programs to ensure the long term sustainability of Canada’s national statistical system," said Mann in his statement. "Our most critical programs such as the census of population, the consumer price index, gross domestic product, trade, and employment statistics continue without interruption."

Still, fewer statisticians means fewer people gathering and crunching numbers, and without them, less certainty in the world around us.

One would think that having a solid grip on things would be a necessity for a nation trying to reinvent itself in the midst of a blistering trade war and a changing world, but that doesn’t seem to be the priority. The wizards must do their best with what they have. In the meantime, the revisions will continue, and reality will warp and blur.

More For You

Our newsletter is (much) better than this pop up

Plus: signing up means you'll never see this pop-up again. Score!