These economists are worried Canada has a data quality problem

Instead, the economy posted a second consecutive quarterly decline — sparking talk of a technical recession.

While most economists, including at the central bank, have dismissed the recession label as premature, those at Desjardins took the latest economic surprise as an excuse to delve into recent volatility in the data.

Desjardins economists concluded in a report published Thursday that Canada faces a “data quality crisis.”

Randall Bartlett, the report’s co-author and deputy chief economist at Desjardins, said in an interview that concern about the reliability of data in Canada and internationally is an “open secret” in economics circles.

Statistics Canada’s view is that volatility should be expected in the data during periods of economic upheaval, but the agency expects those key measures will smooth out as it receives more regular data.

Bartlett said he fears economic forecasts and government policy decisions will suffer if they’re not based on cogent information.

“If the data quality is extremely poor, even so-called evidence-based decisions ultimately could become poor decisions,” he said.

The Desjardins analysis compared the average size of revisions to StatCan’s monthly GDP by industry figures in the pre- and post-pandemic periods.

Desjardins economists found that revisions were much larger on average post-pandemic. Revisions also tended to be bigger when more time had passed since the initial estimates.

A major upward revision to the previous three years of GDP data in November 2025 was one of the factors giving the federal government a rosier fiscal position in the recent spring economic update compared with the fall budget.

There are a number of reasons why StatCan might be revising data more regularly, Bartlett explained.

Much of the agency’s work, such as the monthly labour force statistics, relies on surveys of Canadians. But StatCan and pollsters alike are struggling to get Canadians to answer surveys these days, Bartlett said, limiting these critical sources of data.

Recent changes in federal immigration policy are also severely curtailing population growth, driving large swings that can skew data like employment rates or per-capita metrics.

The monthly labour force survey has been the subject of many regular surprises for economists in recent months, though it’s already considered a notoriously volatile data source.

Not all of the limitations on data are domestic concerns.

StatCan relies on data from international partners to fill out some of its regular reports — nowhere more than on the trade front.

The agency typically exchanges customs data with counterparts in the United States to reconcile trade volumes, but the tariff dispute between the nations and other disruptions south of the border have compromised that flow of information.

“Imports and exports from the U.S. to Canada should mirror those same transactions in the U.S.,” Bartlett said.

“And if there’s less accuracy in terms of the collection and estimation of those numbers south of the border, that has material implications for how we understand our trade with the U.S. here at home.”

When the government shutdown in the United States this past fall, it shuttered the U.S. Census Bureau, and StatCan delayed a pair of scheduled releases for international trade.

When it came time to report third quarter GDP results that rely on those figures, StatCan substituted the missing data with special estimates based on modelling forecasts and some available preliminary data.

Bartlett said that with first-hand, hard data increasingly unavailable to StatCan, the agency has had to rely on projections like these more often to fill the gaps. But that data is more subject to revisions and may be contradicted when statisticians receive official sources.

Carter Mann, chief of media relations at Statistics Canada, said in a statement to The Canadian Press that the agency makes stakeholders aware when situations like the U.S. government shutdown could lead to “larger-than-normal revisions.”

Updates to key macroeconomic indicators after getting new sources of data are a part of regular operations at StatCan, he said.

The agency also does not adjust its modelling to account for structural changes like shifts in population trends or trade disruptions.

“During times of economic volatility, it is expected to see increased volatility in the source data, which overtime will calm once more data receipts are incorporated,” Mann said.

“We do not change our estimation methodology in times of economic volatility. Methodological consistency is important when measuring the economy.”

StatCan is also facing pressures to its operations from the federal government.

Desjardins pointed out in its report that StatCan, like other federal agencies, is expected to find millions of dollars in savings over the coming years as part of the government-wide expenditure review.

Budget 2025 also instructs the agency to increase its efficiency while reducing costs.

“For example, the frequency of data collection will be reduced where the requirements can be met through statistical modelling or other modern methods,” the document reads.

Bartlett said leaning into new tools like artificial intelligence could help improve the accuracy of StatCan’s projections. But he also argued that the agency needs support from the federal government to accurately show what’s happening on the ground in the economy.

“There is certainly room for efficiency gains at Statistics Canada, but ensuring they have the adequate resources they need to provide the best quality data for Canadians and for policymakers is paramount to ensuring we have at least the same starting point for making policy decisions,” Bartlett said.

The Canadian Press has reached out to the office of Industry Minister Mélanie Joly, whose mandate encompasses StatCan, for comment but has not received a response yet.

This report by The Canadian Press was first published June 25, 2026.

Craig Lord, The Canadian Press

by The Canadian Press

These economists are worried Canada has a data quality problem

What's Trending