Rose Roberts shops with her daughter Cerys Mochan, 3, riding in the shopping cart, at a Guelph, Ont. grocery store in this Jan. 17, 2007 photo.DAVE CARTER
Government statisticians who have the important task of closely monitoring inflation face a growing outcry in country after country for seeming to underestimate consumer price increases and their impact on household incomes and spending - either through deliberate policy (Hello Argentina), inaccuracy or a lack of effective tools. So it is more than a little surprising to find a new study from the respected C.D. Howe Institute criticizing Statistics Canada for consistently overestimating consumer price inflation.
Statscan's measurement is too high by about 0.6 percentage points on average, says the study by economist Christopher Ragan - the same conclusion reached in a 2005 Bank of Canada study. "So when the measured inflation rate is at the Bank of Canada's 2 per cent target, the true rate of inflation is closer to 1.4 per cent."
The flawed measurements stem from the CPI's failure to adjust the weights attached to items in its consumption basket as quickly as needed to respond to consumers' frequent shifts in spending patterns. Instead, Statscan relies on past behaviour and makes changes only occasionally, the study says. Weightings within the basket are changed every four years.
These flaws are far from inconsequential, the study argues. They affect everything from crucial monetary and fiscal decisions to income tax collections and pension or other spending tied in some way to the CPI. The report pegs the cost to Ottawa at hundreds of millions of dollars annually.
The agency acknowledges that the weightings need to be adjusted more frequently to better capture consumer patterns. It is currently working with 2005 basket weightings, and will switch to 2009 weights in June.
"We are working to improve that … possibly go to a three-year or even a two-year interval after that," said Richard Evans, director of Statistics Canada's consumer prices division. The agency also intends to speed up the changeover process. "The quicker you can introduce the new weighting pattern, the better it is. If we could do it in a year, that would be an improvement."
The C.D. Howe study singles out a particular flaw in the data gathering: The CPI does not take notice of consumers' natural tendency to substitute cheaper alternatives for products rising in price - known as substitution bias. So the official inflation estimate accords too much weight to products whose prices are rising. This makes the rate of increase higher than what is actually experienced by the average consumer in the marketplace.
Statistics Canada's own calculation of the effect of substitution bias concluded that it adds between 0.15 and 0.2 percentage points to CPI. "Other sources of bias are more difficult to pin down," Mr. Evans said.
These include new products that have not yet been introduced into the CPI; substantial changes in product quality or features; and the effect of new outlets in the marketplace that are not yet adequately represented in the index calculations. If, as is typical, major new entrants have lower pricing than established ones, the index might fail to capture some of that change.
As a result, it's important to introduce the new stores into the sample as soon it can be established that they are large enough, representative enough and attract enough consumer traffic to affect pricing, Mr. Evans said.
Economists agree "that there is upward bias coming from those sources. The challenge is putting a number on it," he said. "There are different ways of measuring them. Different academics using different assumptions have come up with different measurements."
It is not only central banks and finance departments that need as accurate an inflation gauge as possible as they chart the direction of interest-rate policy and determine the costs of inflation-indexed social programs and the threshold income levels for the various tax brackets. Consumers also use the CPI to assess their financial needs and spending capacity.
Now, most consumers would argue forcefully that their costs are heading in one direction - up. But that's because our focus tends to be on such frequently purchased items as food and gasoline. The stuff we pay for much less frequently - clothes, say, or electronics, autos, furniture and apartment rents - are likely to be much less volatile, declining in price or rising much more gradually.
An investigation of U.S. inflation data-gathering methods in the 1990s found an upward bias in the U.S. CPI of 1.1 to 1.3 per cent. At the time, the U.S. Department of Labor was updating its basket of goods only once every 10 years. It subsequently began making adjustments every two years, after the government poured resources into the department.
The solution to the Canadian CPI flaws lies in a relatively small increase in funding to enable Statistics Canada to sift through more consumer price data on a more regular basis and in much finer detail, the study says.
The benefits? "Canadians would have a truer sense of changes in the cost of living; monetary policy would be guided by a more accurate measure of inflation, and [Finance]Minister [Jim]Flaherty would more easily achieve the government's commitment to balance the federal budget by 2015-16."