StatsCan admits past CPI data flawed

By Steven Lamb | August 16, 2006 | Last updated on August 16, 2006
2 min read

Statistics Canada has revealed that it has been miscalculating the consumer price index for the past five years, understating inflation on average by 0.1% since early 2001. The understatement stems from a problem in the formula used to calculate price increases for hotel accommodation. Since 2001, StatsCan has calculated this element of the overall CPI to have fallen by 16%, when it actually rose 32%.

At least one economist says the effects will be limited.

“Although one doesn’t want outright mistakes, the bigger picture is that inflation is only measured approximately — it may in fact overstate the cost of living for a number of reasons,” says Avery Shenfeld, senior economist at CIBC World Markets. “No economic measure is reported with precision, but one does want to avoid creating errors beyond those inherent in measuring inflation.”

He says investors holding real return instruments and workers whose earnings are indexed to inflation should not hold their breath waiting for any remedy, though.

“There will be some who will be disappointed that they were ‘cheated’ out of a small payment that they should have received,” he says. “The CPI is not going to be revised. It’s water under the bridge at this point.”

People currently receiving pension benefits may fare a little better, but only if StatsCan factors the miscalculation into its next CPI report and adjusts the official rate of inflation to reflect the error.

“If they try to incorporate or reflect the miscalculation in the next CPI figures, there might be an impact, but we can’t judge what that would be until we know if and by how they are going to readjust the number,” says Karen DeBortoli, director of the Canadian research and innovation centre at Watson Wyatt Worldwide. “There is currently no information about that that I have seen.”

Even then, any adjustment would be minimal, according to Steve Bonnar, principal of Towers Perrin.

“Those pension plans that provide automatic increases based on inflation will just do a one time catch-up next time around to reflect the error,” he says. “Those sponsors that provide ad hoc increases periodically round so generously that a difference of 0.1% is just not going to be very meaningful.”

Earlier this year, StatsCan admitted it had miscalculated national productivity. That error was traced back to an anomaly in the number of working days in 2005. Still, Shenfeld says StatsCan’s reputation is pretty solid.

“To its credit, Statistics Canada on all of these occasions has admitted its error as opposed to masking a change in a revision,” says Shenfeld. “One never knows if other statistical agencies are as forthcoming or if they present corrected data as ‘revised’ and not necessarily conceding that an error was made. Virtually everyone, at some point makes a mistake.

“Their reputation is quite solid and I don’t think this will seriously sully their reputation.”

Filed by Steven Lamb,,


Steven Lamb