The consumer price index (CPI) may be an imperfect measure of changes in the cost of living owing to measurement biases known as commodity substitution bias, new goods bias, quality bias and outlet substitution bias. When the sum of these individual biases is positive, the rate of change in the CPI overstates the increase in the cost of living. This study attempts to quantify the four potential sources of measurement bias in the Canadian CPI.

Commodity substitution bias arises from the use of fixed weights for individual goods and services in the CPI basket. New goods bias occurs if the prices of new products excluded from the current CPI basket change at a different rate than prices of goods included in the basket. Since the CPI is intended to be a measure of pure price movements, quality bias exists if the prices used to construct the CPI are not adjusted to remove the effect of changes in product quality. Another potential source of measurement error, outlet substitution bias, enters the CPI if there are shifts in the market shares of different types of retailers and if quality-adjusted prices at all types of outlets are not equal.

The upper limit for the total annual bias in the CPI is estimated to be 0.5 per cent. Because most of the judgments incorporated in this estimate are quite generous, however, the total annual bias is likely to be less than 0.5 per cent.