Consumer Price Index (CPI)
Detailed information for September 2002
The Consumer Price Index (CPI) is an indicator of changes in consumer prices experienced by Canadians. It is obtained by comparing through time, the cost of a fixed basket of commodities purchased by consumers.
Data release - October 23, 2002
The Consumer Price Index (CPI) is an indicator of changes in consumer prices experienced by Canadians. It is obtained by comparing through time, the cost of a fixed basket of commodities purchased by consumers. Since the basket contains commodities of unchanging or equivalent quantity and quality, the index reflects only pure price movements.
Between 1913 and 1926 both the Department of Labour and the Dominion Bureau of Statistics (Statistics Canada) each produced separate CPI's, the former using a highly restrictive target population and the latter using the total population of Canada as the target. In the early 1920's, the Dominion Bureau of Statistics used the data collected by the Department of Labour to construct a monthly series beginning in January 1914. The present CPI series are descended from that set of indexes. Some component indexes in the CPI begin with the date they were introduced into the CPI. Data series at the level of urban centres were introduced in January 1971; province-level data series exist since September 1978.
The CPI is widely used as an indicator of the change in the general level of consumer prices or the rate of inflation. Since the purchasing power of money is affected by changes in prices, the CPI is useful to virtually all Canadians. Consumers can compare movements in the CPI to changes in their personal income to monitor and evaluate changes in their financial situation.
The CPI is generally used in four ways. First, it is used to escalate a given dollar value over time to preserve the purchasing power of that value. Thus the CPI is used widely to adjust contracted payments such as wages, rents, leases and child or spousal support allowances. Private and public pension programs (including Old Age Security and the Canada Pension Plan), personal income tax deductions, and some government social payments are also escalated using the CPI. Second, the CPI is used as a tool for deflating current dollar estimates to obtain constant dollar estimates that eliminate the effects of price change. Third, the CPI is used in setting and evaluating economic policies. The Bank of Canada uses the CPI, and special aggregates of the CPI, in determining its monetary policies and to measure the success of its policies to contain inflation within its target range. The CPI and its component indexes can be used in assessing public policy impacts, e.g. agricultural policies and food prices. Finally, business analysts and economists use the CPI for economic analysis and research about such questions as the causes and effects of inflation and regional disparities in price movements.
Price movements of the goods and services represented in the CPI are weighted according to the relative importance of commodities in the total expenditures of consumers. These weights ensure that a 10% price increase in rent, for example, would have a greater impact on the All-items CPI than a 10% increase in the price of milk. This is because Canadians as a whole spend a much larger share of their total expenditures on rent than on milk. CPI basket weights are updated at intervals of about four years; they are obtained from surveys of consumer expenditures.
- Consumer price indexes
- Prices and price indexes
Data sources and methodology
Two populations must be defined, the target population whose expenditures are the basis of CPI weights and the universe of consumer goods and services to be represented by the CPI.
Price collectors collect most price data using technical descriptions or specifications for the selected goods and services in the CPI sample. These are referred to as representative products.
The specification for a representative product contains:
(1) a description (e.g., "frozen French fried potatoes, regular, crinkle or shoe string cut");
(2) acceptable units of measure;
(3) frequency of pricing (e.g., monthly);
(4) quality requirements (e.g. "Grade A or Fancy"); and
(5) description of acceptable varieties (e.g., "super fries, super chips, dollar chips").
Representative products in the price survey may represent larger groups of similar products. Market intelligence and research by CPI product specialists are used to select and define representative products. As much as possible, interviewers track the price of the same products through time. However, a review of sales shares is conducted periodically to ensure that the sample of products continues to be representative. If a product fails to meet sales-share criteria, it is removed from the survey and replaced by another product with higher sales.
This is a sample survey with a longitudinal design.
Samples in the price survey are obtained from a judgmental selection of geographical areas, representative commodities, and types and locations of retail outlets. The timing of price collection during the month is predefined. The extent and quality of the sample is constrained by budgets and the information available on which to base the sampling process. The sample selection process attempts to maximize the relevance and quantity of information.
Annual reviews and updates of product samples ensure that price change based on samples of products are representative of the price change for the entire class of products. New products are introduced into the pricing samples based on market share significance, market and product trends and the judgment of commodity officers. Particular efforts are made to introduce new products into samples for product categories with relatively high rates of change. Outlet sample reviews and updates are performed annually for many CPI basic classes. All outlet samples are reviewed at least every 2 to 3 years.
There are about 600 commodities specified to represent the price movement in 182 basic commodity classes. Sample commodities are chosen on the basis of representativeness and expected continuous availability.
The geographical distribution of the sample varies by commodity. The most geographically dispersed price samples are for commodities whose prices are likely to be heavily influenced by local conditions i.e. locally determined prices for commodities such as water, local transit, rents and property taxes. In contrast, prices of commodities such as car registrations or postage are collected from provincial or national agencies.
The selection of outlets is based on market intelligence and is designed primarily to include retail outlets with high sales revenues. Almost all prices are collected from retail outlets or from local, regional or provincial agencies. However, rent quotes are collected from a survey of tenants, based on the framework of the Labour Force Survey.
The timing of price collection (e.g. the first week of the month) is predefined to ensure consistency. For some commodities whose prices are more variable, price collection is spread over several weeks.
The All-items index at the Canada level is based on an annual sample of over 500,000 price quotes.
Responding to this survey is mandatory.
Data are collected directly from survey respondents, extracted from administrative files and derived from other Statistics Canada surveys and/or other sources.
The CPI collects prices from a price survey and other sources. The price survey collects prices of a large, representative set of consumer goods and services. The frequency of collection of prices of any particular good or service varies depending on the nature of the commodity.
The prices of most of the commodities surveyed for the CPI are collected monthly, usually in the first three weeks of the reference month. Gasoline prices are collected in four weeks of the month.
The prices of some goods are observed less frequently than monthly since their prices tend to change less often. Prices of cameras are observed five times a year. Bus, train and taxi fares are collected twice a year. University tuition fees, property taxes and automobile registration fees are collected once a year. However, when prices change outside of the scheduled time of collection, a special price collection may be carried out. Prices of seasonal goods (e.g. winter coats), which appear in the market place for only a few months of the year, are observed only in the months in which supplies are judged reasonable.
Over 90 percent of the price quotes used in the construction of the CPI are collected by personal visits to selected retail outlets. For products such as newspapers, cable television services, men's haircuts, fuel oil, and restaurant meals, prices are collected by telephone and at least once a year, by personal visits. Some types of information are collected via the Internet.
Price collectors or interviewers in the Regional Offices of Statistics Canada are trained and equipped with hand-held computers containing instructions, technical descriptions of the products to be observed and forms for entering and qualifying the information required. Interviewers visit chosen outlets during a selected period to record the specified data.
Price survey instructions consist of a sample of representative commodities whose prices are observed in selected areas of a city or province, in selected retail outlets, on a specified week of the month. Detailed rules and conventions govern price collection. These rules help ensure that the practical aspects of compiling the CPI are consistent and adhere to the underlying concepts and definitions.
Estimated percentage shares (by weight in the CPI) of the collection methods used in the construction of the CPI:
Computer Assisted Personal Interview (CAPI) 65%
Data derived from other Statistics Canada programs 15%
Telephone interviews 10%
Extraction of data from other sources 10%
Note: The CAPI category includes some paper forms completed by interviewers.
During the process of collecting price data, interviewers are asked to provide the reasons for large or unusual price changes. Edit checks are also done on the hand-held computers at the time of data capture. The price information transmitted to Statistics Canada is further reviewed for outliers or unusual changes. A combination of judgment and outlier detection techniques are used to detect errors. In cases where unusual price changes are not explained, follow up investigations are made. Product specialists in the Consumer Prices Division also prepare monthly reports summarizing and explaining price movements for products in the survey.
Imputations can be used in the CPI for several reasons. First, not all commodities covered in the price survey are observed on a monthly frequency. Among these are seasonal goods available for only part of the year (e.g. lawn mowers), goods whose prices are relatively stable (e.g. parking), and some services whose prices change only once a year (e.g. university tuition fees). Second, the prices of some commodities are not collected in some geographical areas mostly because the quantities available to consumers in those areas are too limited to merit the cost of collecting price data (e.g. fuel oil in Alberta). Third, prices may be imputed in cases where prices are missing in a given month due to out of stock or otherwise not available, or because the prices collected are questionable. Finally, adjustments to price data must be made in cases of quality change.
Imputation rules identify an imputation source and target. The source is from where the price change information is derived and the target is the commodity or geographical area to which that information is applied.
For seasonal goods, in their out-of-season months, price movements are obtained from the larger group to which the seasonal goods belong. For goods with relatively stable prices, an assumption of no price change is used in the months for which prices are not collected. Similarly for commodities whose prices change once a year, no price change is assumed in non-priced months. In the latter two cases, index levels remain unchanged when prices are not observed.
In geographical imputations, price movements in other areas are used as source to substitute for price movements in places where prices are not observable, for example, the use of BC fuel oil prices for Alberta.
To compare prices that relate to different quantities, prices are adjusted to a standard quantity unit. When an item is temporarily out of stock, the last recorded price is retained unless that price was a sale price, in which case the last regular price is used. If an item is no longer available at a given outlet, the item is replaced by a similar one in the same outlet or, if necessary, in a different outlet.
In cases where an explicit adjustment for quality change is impractical, prices are not directly compared. In these cases, a new item is introduced into the sample for which price movements can only be calculated in the following month.
Estimation procedures are used to derive weights in the CPI basket. The main sources of expenditure data on consumer goods and services are the consumer expenditure surveys (the Family Expenditure Survey up until 1996, now called the Survey of Household Spending and the Food Expenditure Survey). Average yearly expenditures per household are calculated for each commodity class by province or sub-provincial area. These are then applied to the estimated number of households in each geographical area giving aggregate expenditures for each commodity class. Aggregate expenditures for Canada are obtained from estimated aggregate expenditures for each basic commodity group for each geographical unit.
A step-by-step approach is used to calculate the fixed-basket consumer price indexes for any aggregate above the basic class level. First, indexes for basic classes -in which cases price movements are measured as the ratio of their prices in an observed period to their prices in the base period- are combined (averaged) into the next-level composite indexes, these in turn into higher-level indexes, etc., up to the level of "All items". Ultimately, though, every composite index is a weighed arithmetic average of the corresponding price indexes for all of the basic classes contained in the given aggregate.
Generally, the factors affecting the quality of the CPI include:
- the size and composition of the price samples of goods and services and outlets;
- the accuracy of the expenditure estimates used to assign weights;
- the frequency and speed of updating of the contents and weights of the CPI basket;
- the effectiveness of error detection and correction, and imputation methods for missing data; and
- the application of appropriate methods of adjusting for quality change of goods and services in the CPI sample.
The quality of this index is maintained through the expertise of the analysts assigned to it. They develop a thorough knowledge of the domain, which is supplemented by outside personal contacts for particular goods or services. Much time and effort is devoted to detecting and following up on unusual fluctuations over time in the pricing patterns of goods and services. Prior to dissemination, the price indexes are analyzed and historic trends reviewed.
Statistics Canada is prohibited by law from releasing any information it collects which could identify any person, business, or organization, unless consent has been given by the respondent or as permitted by the Statistics Act. Various confidentiality rules are applied to all data that are released or published to prevent the publication or disclosure of any information deemed confidential. If necessary, data are suppressed to prevent direct or residual disclosure of identifiable data.
Collected data are converted to price indexes and data are released as such, so that it is not possible to identify the suppliers of raw prices.
Revisions and seasonal adjustment
The CPI allows to compare, in percentage terms, prices in any given time period to prices in the official base period which, at present is 1992=100. The official time base was changed from 1986=100 to 1992=100 starting with the CPI for January 1998. The change is strictly an arithmetic conversion, which alters the index levels but leaves the percentage changes between any two periods intact, except for differences in rounding.
The seasonally adjusted All-items CPI for Canada by month is obtained by aggregating seasonally adjusted data for the 8 major component groups. Seasonally adjusted series are also produced for several special aggregate indexes at the Canada level: "All-items excluding Food", "All-items excluding Food and Energy", and "Energy" indexes. More recently a seasonally adjusted special aggregate index was added, "All-items excluding the eight most volatile components, as defined by the Bank of Canada". Seasonally adjusted index data are available on CANSIM, the electronic database of Statistics Canada.
The seasonal adjustment process uses Statistics Canada's X-11 ARIMA program. Seasonally adjusted data are revised once a year with the release of the indexes for January.
Statistical reliability is inherently more difficult to assess for price indexes than for other statistical series due to the complex nature of composite price change and the statistical problems of estimating composite price change.
The quality of the CPI depends on many factors related to price and consumer expenditure. Data for both price change and consumer expenditure patterns are obtained from sample surveys and therefore are subject to sampling error. Generally, the accuracy of price data is more important to overall quality of the CPI than the accuracy of expenditure data used for weights.
Sampling error for the CPI price surveys is difficult to quantify because, except for rents and traveller accommodation, product and outlet samples are based on judgmental rather than probability sampling methods.
Several studies related to aspects of data quality for the CPI are available. All of these papers are available free in "Analytical series -Prices Division", Statistics Canada, catalogue no. 62F0014M, which can be accessed online from the "Browse by key resource" module of the Statistics Canada website under "Publications".
As well, the report from an assessment of data quality of the Canadian CPI performed by the International Monetary Fund in 2003 is available on the IMF website at the following URL:
In summary, for each of the 15 applicable elements assessed, the CPI was rated at the highest of a four-level data quality assessment grid.
The CPI is believed to be sufficiently accurate for most practical purposes. Accuracy is best at higher levels of geographic and product aggregation due to the larger sizes of the price samples for high levels of aggregation. As well, since errors can occur in price collection and editing, or in making quality adjustments to estimate "pure price change", higher level aggregate indexes are likely to be of better quality than lower level indexes because any distortions due to errors are more likely to cancel out. In general, accuracy is better at the Canada level for any product index in the CPI, compared to the same index at the province or city level. Also, accuracy is better at the All-items CPI or major component levels of the CPI, compared to individual product indexes. Finally, the CPI is more accurate as an indicator of change over several months or a year, compared to the accuracy of the price change measured from any one month to the next.
- The Consumer Price Index - Technical Notes - August 2001
These technical notes are valid from August 2001 to December 2002.
- Weights for the Consumer Price Index
- The Consumer Price Index Reference Paper
Update based on 1992 Expenditures, Statistics Canada, Catalogue 62-553, Occasional
- Changes to the Consumer Price Index beginning with the January 2003 CPI (Product detail availability in monthly publication and CANSIM II)