StatCan fails to capture high food prices
Experts suggest blame lies with sourcing data from just three large grocery chains
Every week, Melanie Morrison gathers data on the prices charged by Canadian grocers for tens of thousands of products.
But these days, Morrison, who runs a company based in Saskatoon called BetterCart Analytics, is mostly interested in the price of just three of them: condensed milk, peanut butter and plain old butter.
Based on the data she’s collected on these three products, Morrison is convinced that Statistics Canada’s efforts to track food price inflation data may require a big rethink.
Poor data collection on food prices means Canadians may be finding it much harder to put food on the table than the federal government is acknowledging, after all.
Take the federal agency’s own findings, Morrison said. In September 2019, Canadians paid an average of $2.82 for 500 grams of peanut butter, according to Statistics Canada.
In September 2020, that price had dropped to $2.69. And by September 2021, it was back at $2.82.
“I don’t see this in my data,” said Morrison. “We see a range from $3.39 to $5.59 in the current price. And at the highest, we see up to $9.99.”
And then there’s butter. In Morrison’s estimation, Statistics Canada’s data showing that the average price of 454 grams of butter has grown 5.9 per cent over the past year is terribly off.
“Our data indicates the real rate of butter price inflation over the past year has been 19.8 per cent on average across Canada,” Morrison said.
“I’m concerned about the validity of quite a lot of this data.”
She has reason to be: millions of Canadians are living on tight food budgets and low fixed incomes that are pegged to the rate of inflation for living costs, including the inflation of prices for food staples like butter, cooking oil, meat and potatoes.
And it’s not just the food prices and inflation figures that Statistics Canada reports that has Morrison concerned.
She said container sizes aren’t being accurately reflected in some cases, meaning the amount of food for the price may be shrinking.
Take evaporated milk, for instance. Statistics Canada tracks the price of 385-millimetre cans. But those cans don’t exist anymore in Canadian grocery stores.
“It’s now sold in 354-millimetre cans,”
Morrison said. “The cans have shrunk.”
The prices on the other hand, have not. That may mean, due to what’s referred to as “shrinkflation,” that the cost of items like evaporated milk may have gone up in a manner possibly overlooked by Statistics Canada, even though the agency makes efforts to adjust for varying container sizes.
Every month, Statistics Canada offers food price data, as well as data on everything from footwear to automobiles, in a monthly report called the Consumer Price Index. For more than 90 years, the CPI has tracked the prices of scores of Canadian products to help governments and businesses adjust payments to millions of pensioners and social assistance recipients according to prevailing rates of cost inflation.
But Morrison believes Statistics Canada’s food price monitoring efforts, and its 4.2 per cent annual food cost inflation figure for the past year, may be based on data that’s somewhat too far removed from Canadian shops and kitchens. If the government isn’t accurately tracking the inflating cost of something as basic as butter, how reliable is the rest of the CPI?
Until recently, Statistics Canada had a nearly total monopoly on the data needed to calculate food price inflation rates.
But the internet has changed that, said Morrison, whose company claims to be able to track real-world food prices more comprehensively than Statistics Canada to help grocers make what it calls “datadriven decisions in real time.”
Morrison’s cornucopia of grocery price data is gathered from thousands of grocery stores — chains and independents alike — across the country, using online sources such as grocery store flyers and their websites.
It’s a complex business, propelled by massive computing power involving “web-scraping” and data-mining tasks. She now has close to a billion products recorded in her databases.
Not all of Morrison’s clients are grocers with cutthroat price competition in mind, however.
She also collaborates with Sylvain Charlebois, director of Dalhousie University’s AgriFood Analytics Lab in Halifax. Together, they’re probing the accuracy of the CPI’s food price inflation figures and Statistics Canada’s data gathering when it comes to food prices.
Charlebois and Morrison both emphasized that Statistics Canada has “a stellar reputation” when measuring macroeconomic metrics for policy-makers, industry and consumers alike.
The CPI, said Charlebois, is “likely accurate about many aspects of our economy,” including durable goods, automobiles, energy and lodging.”
But food distribution, he said, is “becoming more complicated as market dynamics are much more intense than they used to be.”
In Charlebois’s opinion, Statistics Canada’s biggest problem in tracking food prices is that it likely relies too heavily on just three national chains within Canada’s highly concentrated grocery industry for much of its price data.
Unlike in the U.S., where the Department of Agriculture’s Economic Research Service relies in large part on a vast consumer network for food price data from 120,000 households, Statistics Canada largely relies on data from the three chains it “partners” with. Due to “corporate confidentiality” Statistics Canada can’t disclose the names of the three chains, said Heidi Ertl, director of consumer prices at StatCan.
“We have oligopolistic conditions in the grocery industry,” said Charlebois, with reference to the very small number of chains that control about 80 per cent of the grocery market. With just three companies providing so much of Ottawa’s food price data, “I suspect Statistics Canada’s methodologies may be a victim of that,” he said.
Charlebois also wonders if Statistics Canada is doing enough to track prices in Canada’s independent groceries.
These questions feed his basic concern that Ottawa underestimates budget pressures on lowincome Canadian families, especially pensioners and social assistance recipients whose incomes are often pegged to the CPI’s stated rate of inflation.
Contacted via phone and email, Jason McLinton, vicepresident of grocery division and regulatory affairs for the Retail Council of Canada (RCC), declined to discuss whether “oligopolistic” market conditions exist within the Canadian retail food industry, and whether — in part because of this — the retail price data that Statistics Canada receives from grocery retailers does not sufficiently represent the actual prices Canadians pay for food.
“It is not surprising that results from Statistics Canada’s report, whose data and methodology are easily accessible online, might differ from an investigation whose data and methodology are not,” McLinton said in response to questions about Charlebois’ investigation with BetterCart.
Charlebois, however, noted that Statistics Canada’s food price inflation calculations are based on raw data from grocers that’s not available online and not at all otherwise easily accessible to the public or independently verifiable.
It’s gathered using a data collection process that he said is not explained in anything like sufficient detail to be considered publicly transparent. It’s “very ambiguous,” he said. As a whole, Charlebois gives Statistics Canada rather low marks for public disclosure concerning its raw data and data-gathering methodologies for the food section of the CPI, although he senses that may now be changing as public dismay over food price inflation grows.
Of Canada’s five main grocery chains — Loblaw, Metro, Sobeys, Pattison and Walmart — only Metro responded to interview requests on the topic of its role in helping Statistics Canada track food prices.
But it wasn’t because Metro wanted to discuss the topic. “I’m sorry, but we don’t have someone who can speak on this topic,” Metro communications manager Stephanie Bonk said in an email.
Statistics Canada’s Ertl said she considers Charlebois’s criticisms of the way food prices are documented unfair and incorrect.
Prior to the COVID-19 pandemic, food prices were collected by interviewers visiting stores, she said, adding “this is now replaced by online prices using store websites or flyers.”
Beyond this, she explained, food prices are now mostly captured using weekly scanner data collected directly from the three unnamed grocery chains.
“The data are collected on a weekly basis and include sales and promotions where applicable,” Ertl said.
Food prices are collected from about 500 grocery outlets across all regions in Canada, and “17,000 target products are captured from grocery retailers,” she added.
Ertl acknowledged that Statistics Canada may not be getting enough data, however, adding that she has been lobbying to persuade other major national grocery chains to participate in the data gathering.
“We’re close to reaching an arrangement with a fourth chain,” she said. “We’re trying to cover nationwide. We’d love to get retail scanner data from as many as we can.”
That’s good news, said Charlebois.
But, in an era when a grocery superstore carries as many as 40,000 food products, tracking 17,000 products, impressive as that sounds, may not be good enough, he said, given that Statistics Canada ought to be able to harness the computing power needed to do so.
Ertl said a CPI modernization program has been underway since 2018 to overhaul its methodologies and expand its datagathering reach.
“There’s always room for improvement” she said.
Back in Saskatoon, Morrison said she’s keen to help Ertl in any way she can. “We’re not looking for a fight with StatCan. In an ideal world, we’d be able to help them.”